Drew Brandsen Robert Hoff Spencer Olson Jared Zoodsma

Drew Brandsen Robert Hoff Spencer Olson Jared Zoodsma
Drew Brandsen
Robert Hoff
Spencer Olson
Jared Zoodsma
Engineering 339/340 Senior Design Project
Calvin College
May 15, 2014
© 2014, Drew Brandsen, Robert Hoff, Spencer Olson, Jared Zoodsma & Calvin College
EXECUTIVE SUMMARY
Security is paramount for presidential motorcades, military convoys, and other safety critical
transport. Visibility is often impaired for those inside the vehicles, making it difficult to see all
possible threats ahead, behind, and to the sides. Lack of visibility creates a significant danger from
insurgents, rocket propelled grenades (RPGs), improvised explosive devices (IEDs), and other
threats. Moreover, travel routes can span hundreds of miles. Covering these routes with explosive
detectors, bomb-sniffing dogs, and law enforcement is very costly and requires extra labor.
Improving visibility for individuals in the vehicles can help mitigate these external risks
The solution our team of four electrical and computer engineers has chosen consists of two
subsystems: one to fly above the vehicle and a second installed inside the vehicle. We considered
various options for the first subsystem and chose a quadcopter because of its high expandability,
maneuverability, and stability. The quadcopter is equipped with video capture equipment, as well as
a global positioning system (GPS) receiver. Second, a base station located in the vehicle includes a
computer, Wi-Fi router, GPS receiver, radio controlled (RC) transmitter, and RC interface
module. The base station also contains a graphical user interface (GUI) that receives the video
streamed from the quadcopter and provides a way for the user to interact with the system.
The design process entailed building a quadcopter and expanding it to fit the project
specifications. Some of the key specifications were providing a video stream of the vehicle and its
surroundings, being small enough to fit the entire system in the vehicle, and being able to track the
vehicle autonomously. This involved adding both a camera and a Raspberry Pi development board
to the quadcopter in order to capture and send the quadcopter’s GPS location and video stream over
Wi-Fi to the base station below. The base station computer, with the assistance of an Arduino
microcontroller, used the GPS data in order to autonomously control the quadcopter via RC.
The Eagleye team showed the feasibility of this solution through first semester research and
preliminary designs. Following this, the team designed a working prototype in the second semester.
The team operated under a budget of $1455 in order to complete this work. The team hopes that the
work done on this project will increase awareness of unmanned aerial vehicles (UAVs) and their
beneficial applications.
TABLE OF CONTENTS
1
INTRODUCTION .................................................................................................................................1
1.1
CALVIN COLLEGE ENGINEERING DEPARTMENT...............................................................................1
1.2
TEAM MEMBER BIOGRAPHIES .........................................................................................................1
1.2.1 Jared Zoodsma ...........................................................................................................................2
1.2.2 Robert Hoff .................................................................................................................................2
1.2.3 Drew Brandsen...........................................................................................................................2
1.2.4 Spencer Olson ............................................................................................................................ 2
1.3
REPORT FORMAT ............................................................................................................................. 2
1.4
DESIGN NORMS................................................................................................................................ 3
1.4.1 Transparency ............................................................................................................................. 3
1.4.2 Justice.........................................................................................................................................3
1.4.3 Trust ...........................................................................................................................................3
1.5
ACKNOWLEDGEMENTS ....................................................................................................................3
2
PROJECT REQUIREMENTS .............................................................................................................5
2.1
PROBLEM STATEMENT .....................................................................................................................5
2.1.1 Solution ......................................................................................................................................5
2.1.2 Target Customers .......................................................................................................................5
2.2
CUSTOMER REQUIREMENTS .............................................................................................................5
2.2.1 Performance ............................................................................................................................... 5
2.2.2 Physical .....................................................................................................................................9
2.2.3 Customer ...................................................................................................................................9
2.3
DELIVERABLES .............................................................................................................................. 11
3
SYSTEM DESIGN ............................................................................................................................... 12
3.1
DECISION MATRIX ......................................................................................................................... 12
3.1.1 Possible Solutions .................................................................................................................... 12
3.1.2 Decision Criteria and Justification .......................................................................................... 15
3.1.3 Justification .............................................................................................................................. 16
3.1.4 Best Solution ............................................................................................................................ 21
3.2
OVERALL SYSTEM BLOCK DIAGRAM ............................................................................................. 21
3.2.1 Hardware ................................................................................................................................ 21
3.2.2 Software ................................................................................................................................... 23
3.3
FEATURE MATRIX .......................................................................................................................... 25
3.3.1 Feature Descriptions ................................................................................................................ 26
3.4
FEATURE RESULTS......................................................................................................................... 29
3.4.1 Feature Descriptions ................................................................................................................ 29
4
QUADCOPTER DESIGN ................................................................................................................... 31
4.1
QUADCOPTER ARCHITECTURE ....................................................................................................... 31
4.2
QUADCOPTER COMPONENTS .......................................................................................................... 34
4.2.1 Flight Controller ...................................................................................................................... 35
4.2.2 GPS Module ............................................................................................................................. 39
4.2.3 Propellers ................................................................................................................................. 40
4.2.4 Motors ...................................................................................................................................... 41
4.2.5 ESCs ......................................................................................................................................... 42
4.2.6 Battery ...................................................................................................................................... 43
4.2.7 Power Distribution System ....................................................................................................... 46
4.2.8 Frame and Landing Gear ......................................................................................................... 47
4.2.9 VGPU ....................................................................................................................................... 49
4.2.10 Camera ................................................................................................................................... 57
ii
5
BASE STATION DESIGN .................................................................................................................. 59
5.1
BASE STATION ARCHITECTURE ..................................................................................................... 59
5.2
BASE STATION COMPONENTS ........................................................................................................ 62
5.2.1 Base Station Computer ............................................................................................................. 62
5.2.2 RC Interface Module ................................................................................................................ 69
5.2.3 RC Transmitter ......................................................................................................................... 82
5.2.4 GPS Module ............................................................................................................................. 83
5.2.5 Wireless Router ........................................................................................................................ 85
6
SYSTEM INTEGRATION AND TESTING ..................................................................................... 87
6.1
TELEMETRY STREAMING ............................................................................................................... 87
6.1.1 Test: Full System Communication ............................................................................................ 87
6.1.2 Test: Communication with Live Telemetry Coordinates .......................................................... 87
6.1.3 Communication Latency........................................................................................................... 88
6.2
VIDEO STREAMING ........................................................................................................................ 89
6.2.1 Video Latency ........................................................................................................................... 89
6.2.2 Video Quality ........................................................................................................................... 90
6.3
GPS TRACKING ............................................................................................................................. 91
6.3.1 Stationary GPS Precision......................................................................................................... 91
6.3.2 Moving GPS Precision ............................................................................................................. 92
6.4
RC CONTROL VIA RC INTERFACE MODULE ................................................................................... 95
6.4.1 RC Pass Through ..................................................................................................................... 95
6.4.2 Pitch and Roll Recording ......................................................................................................... 95
6.4.3 Steps ......................................................................................................................................... 96
6.4.4 Autonomous Pitch and Roll Start ............................................................................................. 96
6.4.5 Autonomous Pitch Stop ............................................................................................................ 97
6.5
TRACKING ALGORITHM ................................................................................................................. 97
6.5.1 RC Response Test ..................................................................................................................... 97
6.5.2 Motor Response Test ................................................................................................................ 98
6.5.3 North-Facing Flight Tracking .................................................................................................. 99
6.5.4 Orientation Tracking .............................................................................................................. 100
7
PROJECT MANAGEMENT ............................................................................................................ 103
7.1
DOCUMENTATION ORGANIZATION .............................................................................................. 103
7.1.1 Code ....................................................................................................................................... 103
7.1.2 Design Journals ..................................................................................................................... 103
7.1.3 Manuals.................................................................................................................................. 103
7.1.4 Media ..................................................................................................................................... 103
7.1.5 Miscellaneous Notes............................................................................................................... 103
7.1.6 Planning Documents .............................................................................................................. 103
7.1.7 Status Reports ........................................................................................................................ 104
7.1.8 Team 08: Turn-in ................................................................................................................... 104
7.1.9 Tests ....................................................................................................................................... 104
7.2
TEAM ORGANIZATION ................................................................................................................. 104
7.2.1 Technical Assignments (Design Areas) .................................................................................. 104
7.2.2 Management Assignments ...................................................................................................... 106
7.2.3 Working Guidelines ................................................................................................................ 106
7.2.4 Safety Guidelines .................................................................................................................... 107
7.2.5 Team Communication and Accountability ............................................................................. 107
7.3
SCHEDULE AND WORK BREAKDOWN SCHEDULE ......................................................................... 107
7.3.1 Hours Summary ...................................................................................................................... 107
7.3.2 Milestones .............................................................................................................................. 108
7.4
OPERATIONAL BUDGET ............................................................................................................... 109
7.4.1 Project Costs .......................................................................................................................... 109
iii
7.4.2 Summary................................................................................................................................. 110
7.4.3 Sources of Funding ................................................................................................................ 110
7.5
METHOD OF APPROACH ............................................................................................................... 110
7.5.1 Design Methodology .............................................................................................................. 110
7.5.2 Research Techniques .............................................................................................................. 110
8
BUSINESS PLAN .............................................................................................................................. 111
8.1
OVERVIEW ................................................................................................................................... 111
8.2
VISION AND MISSION STATEMENT ............................................................................................... 111
8.3
VISION FOR THE COMPANY .......................................................................................................... 111
8.4
VALUES AND PRINCIPLES ON WHICH THE BUSINESS STANDS ....................................................... 111
8.5
MARKETING STUDY ..................................................................................................................... 111
8.5.1 How Large is the Market? ...................................................................................................... 111
8.5.2 Is it Growing or Shrinking? ................................................................................................... 112
8.5.3 Regulatory Restrictions .......................................................................................................... 112
8.5.4 Barriers to Entry and Exit ...................................................................................................... 112
8.5.5 Key Success Factors in the Industry....................................................................................... 112
8.6
COMPETITION .............................................................................................................................. 112
8.6.1 Existing Competitors .............................................................................................................. 112
8.6.2 Potential Competitors ............................................................................................................ 113
8.7
BUSINESS MODEL ........................................................................................................................ 114
8.7.1 Desired Image and Position in the Market............................................................................. 114
8.7.2 Company Goals and Objectives ............................................................................................. 114
8.7.3 SWOT Analysis ....................................................................................................................... 114
8.7.4 Competitive Strategy .............................................................................................................. 115
8.8
FINANCIAL FORECASTS ................................................................................................................ 116
8.8.1 Key Assumptions .................................................................................................................... 116
8.8.2 Financial Statements .............................................................................................................. 117
8.8.3 Break-Even Analysis .............................................................................................................. 119
8.8.4 Ratio Analysis ........................................................................................................................ 119
8.9
SUMMARY.................................................................................................................................... 120
9
CONCLUSION .................................................................................................................................. 121
9.1
PROJECT RESULTS ....................................................................................................................... 121
9.1.1 Overall ................................................................................................................................... 121
9.1.2 Final Costs ............................................................................................................................. 121
9.1.3 Time Spent .............................................................................................................................. 121
9.2
FUTURE WORK ............................................................................................................................ 121
9.3
SUMMARY.................................................................................................................................... 122
iv
TABLE OF FIGURES
FIGURE 1: TEAM PHOTO - JARED ZOODSMA, DREW BRANDSEN, ROBERT HOFF, AND SPENCER OLSON. ........................1
FIGURE 2: 2-D HORIZONTAL DISTANCE REQUIREMENT FOR PROTOTYPE 1.0 .................................................................6
FIGURE 3: QUADCOPTER OUTFITTED WITH CAMERAS ................................................................................................. 12
FIGURE 4: RC HELICOPTER .......................................................................................................................................... 13
FIGURE 5: RC AIRPLANE .............................................................................................................................................. 13
FIGURE 6: RC BLIMP .................................................................................................................................................... 14
FIGURE 7: PANORAMIC BALL CAMERA ........................................................................................................................ 14
FIGURE 8: CONTROL OF A QUADCOPTER ..................................................................................................................... 18
FIGURE 9: PLATFORM AREA OF A QUADCOPTER .......................................................................................................... 20
FIGURE 10: OVERALL SYSTEM HARDWARE BLOCK DIAGRAM ..................................................................................... 22
FIGURE 11: OVERALL SOFTWARE SYSTEM BLOCK DIAGRAM ...................................................................................... 24
FIGURE 12: QUADCOPTER PROTOTYPE 1.0 ................................................................................................................... 31
FIGURE 13: QUADCOPTER HARDWARE BLOCK DIAGRAM ............................................................................................ 32
FIGURE 14: QUADCOPTER SOFTWARE BLOCK DIAGRAM.............................................................................................. 33
FIGURE 15: CENTRAL QUADCOPTER COMPONENTS ...................................................................................................... 34
FIGURE 16: COMPASS VALUES WITH HEADING-HOLD MODE ....................................................................................... 38
FIGURE 17: PROTOTYPE 1.0 MOTOR DIRECTION .......................................................................................................... 42
FIGURE 18: PROTOTYPE 1.0 FLIGHT TIMES WITH VARIOUS BATTERY SIZES ................................................................. 45
FIGURE 19: BROKEN ARM ............................................................................................................................................ 48
FIGURE 20: PROPELLER GUARDS WITH VIBRATION PROBLEMS (PROTOTYPE 0.2) ........................................................ 48
FIGURE 21: FINAL PROPELLER GUARD IMPLEMENTATION (PROTOTYPE 1.0) ............................................................... 49
FIGURE 22: CUSTOM VGPU LAYOUT ........................................................................................................................... 50
FIGURE 23: VGPU SOFTWARE BLOCK DIAGRAM ......................................................................................................... 52
FIGURE 24: RASPBERRY PI DIGITAL I/O PINS ............................................................................................................... 56
FIGURE 25: MULTIW II SERIAL PROTOCOL ................................................................................................................... 56
FIGURE 26: MULTIW II PRO CONNECTION .................................................................................................................... 57
FIGURE 27: FINAL BASE STATION IMPLEMENTATION ................................................................................................... 59
FIGURE 28: BASE STATION HARDWARE BLOCK DIAGRAM ........................................................................................... 60
FIGURE 29: BASE STATION SOFTWARE BLOCK DIAGRAM ............................................................................................ 61
FIGURE 30: BASE STATION COMPUTER SOFTWARE BLOCK DIAGRAM ......................................................................... 63
FIGURE 31: FINAL GUI ................................................................................................................................................. 65
FIGURE 32: RC INTERFACE MODULE HARDWARE BLOCK DIAGRAM ........................................................................... 69
FIGURE 33: CUSTOM RC INTERFACE MODULE LAYOUT............................................................................................... 71
FIGURE 34: FILTER BOARD CIRCUIT ............................................................................................................................. 74
FIGURE 35: RC INTERFACE MODULE OUTPUT DIAGRAM ............................................................................................. 75
FIGURE 36: FILTER BOARD LAYOUT ............................................................................................................................ 76
FIGURE 37: FILTER BOARD PHOTO ............................................................................................................................... 76
FIGURE 38: RC INTERFACE MODULE SOFTWARE BLOCK DIAGRAM............................................................................. 77
FIGURE 39: DISTANCE DEPENDENT SPEED METHOD .................................................................................................... 80
FIGURE 40: ORIENTATION TRACKING........................................................................................................................... 81
FIGURE 41: RC INTERFACE MODULE EXTERNAL PHOTO .............................................................................................. 84
FIGURE 42: RC INTERFACE MODULE INTERNALS ......................................................................................................... 84
FIGURE 43: VIDEO LATENCY RESULTS ......................................................................................................................... 90
FIGURE 44: VIDEO PROTOCOL QUALITY SCORES ......................................................................................................... 91
FIGURE 45: MOVING GPS PRECISION TEST (22.35 M/S, 10 HZ) .................................................................................... 93
FIGURE 46: MOVING GPS PRECISION TEST (16.9 M/S, 10 HZ) ...................................................................................... 94
FIGURE 47: ORIENTATION FLIGHT TEST ..................................................................................................................... 101
FIGURE 48: GRAPH OF HOURS WORKED ON PROJECT................................................................................................. 108
v
TABLE OF TABLES
TABLE 1: PROJECT DELIVERABLES ............................................................................................................................... 11
TABLE 2: DECISION MATRIX ........................................................................................................................................ 15
TABLE 3: SYSTEM HARDWARE BLOCK DIAGRAM INTERFACE DESCRIPTIONS .............................................................. 22
TABLE 4: SYSTEM SOFTWARE BLOCK DIAGRAM INTERFACE DESCRIPTIONS ............................................................... 24
TABLE 5: SYSTEM PROTOCOL DEFINITIONS ................................................................................................................. 25
TABLE 6: FEATURE MATRIX ......................................................................................................................................... 25
TABLE 7: FEATURE MATRIX IMPLEMENTATION ........................................................................................................... 29
TABLE 8: QUADCOPTER HARDWARE BLOCK DIAGRAM INTERFACE DESCRIPTIONS ..................................................... 32
TABLE 9: QUADCOPTER SOFTWARE BLOCK DIAGRAM INTERFACE DESCRIPTIONS ...................................................... 34
TABLE 10: FLIGHT CONTROLLER ALTERNATIVES ........................................................................................................ 35
TABLE 11: ALTITUDE-HOLD TESTS WITH CONTROL AND FOAM COVERS (METERS) ..................................................... 36
TABLE 12: BATTERY COMPARISONS............................................................................................................................. 44
TABLE 13: BATTERY CAPACITIES AND FLIGHT TIMES .................................................................................................. 45
TABLE 14: COMPONENT POWER NEEDS ....................................................................................................................... 46
TABLE 15: VGPU SOFTWARE BLOCK DIAGRAM INTERFACE DESCRIPTIONS ............................................................... 52
TABLE 16: VGPU PROTOCOLS ..................................................................................................................................... 52
TABLE 17: BASE STATION HARDWARE INTERFACE DESCRIPTIONS .............................................................................. 60
TABLE 18: BASE STATION SOFTWARE BLOCK DIAGRAM INTERFACES ......................................................................... 62
TABLE 19: BASE STATION SOFTWARE PROTOCOLS ...................................................................................................... 62
TABLE 20: BASE STATION COMPUTER SOFTWARE BLOCK DIAGRAM INTERFACES ...................................................... 64
TABLE 21: GUI SOFTWARE PROTOCOLS ...................................................................................................................... 64
TABLE 22: RC INTERFACE MODULE HARDWARE BLOCK DIAGRAM INTERFACES ........................................................ 69
TABLE 23: RC INTERFACE MODULE DEVELOPMENT BOARD ALTERNATIVES .............................................................. 70
TABLE 24: RC INTERFACE MODULE SOFTWARE BLOCK DIAGRAM INTERFACE DESCRIPTIONS ................................... 78
TABLE 25: ANOTHER GPS STATIONARY RESULT (10 HZ) ............................................................................................ 92
TABLE 26: MOVING GPS PRECISION TEST (1 HZ) ........................................................................................................ 93
TABLE 27: MOVING GPS PRECISION TEST (22.35 M/S, 10 HZ) ..................................................................................... 94
TABLE 28: MOVING GPS PRECISION TEST (16.9 M/S, 10 HZ) ....................................................................................... 94
TABLE 29: EXPECTED REACTIONS TO POSITION CHANGES ........................................................................................... 98
TABLE 30: DESIGN AREAS AND TEAM MEMBER ASSIGNMENTS ................................................................................. 105
TABLE 31: SUMMARY OF HOURS ................................................................................................................................ 107
TABLE 32: TEAM 08: EAGLEYE BUDGET .................................................................................................................... 109
TABLE 33: SUPPORTING CALCULATIONS .................................................................................................................... 117
TABLE 34: INCOME STATEMENT ................................................................................................................................. 118
TABLE 35: CASH FLOW STATEMENT .......................................................................................................................... 118
TABLE 36: BREAK-EVEN ANALYSIS ........................................................................................................................... 119
TABLE 37: RATIO ANALYSIS ...................................................................................................................................... 120
vi
TABLE OF ACRONYMS
ABET:
API:
APM:
BEC:
ESC:
FPS:
GPS:
GUI:
HD:
IED:
I/O:
IR:
Kv:
MSP:
MW:
NMEA:
PCB:
PID:
PPFS:
PPS:
PWM:
OS:
Q&A:
RAM:
RC:
RPG:
SD:
SPS:
SSH:
SWOT:
TCP/IP:
U.S.:
UAV:
USB:
VDC:
VGPU:
WBS:
Accreditation Board for Engineering and Technology
Application programming interface
ArduPilot Mega
Battery Eliminator Circuit
Electronic Speed Controller
Frames per Second
Global Positioning System
Graphical User Interface
High Definition
Improvised Explosive Device
Input/Output
Infrared
Revolutions per Minute / Volt
MultiWii Serial Protocol
MultiWii
National Marine Electronics Association
Printed Circuit Board
Proportional-Integral-Derivative
Project Proposal and Feasibility Study
Picture Parameter Set
Pulse Width Modulation
Operating System
Question and Answer
Random Access Memory
Radio Controlled
Rocket Propelled Grenade
Standard Definition
Sequence Parameter Set
Secure Shell
Internal Strengths and Weaknesses, External Opportunities and Threats.
Transmission Control Protocol/Internet Protocol
United States of America
Unmanned Aerial Vehicle
Universal Serial Bus
Volts Direct Current
Video and Graphics Processing Unit
Work Breakdown Structure
vii
1
Introduction
The Calvin College Engineering Program concludes its four-year degree with a Senior Design
Capstone project. This project spans the entire senior year, giving students the opportunity to use
knowledge and skills accumulated throughout their academic careers. It also allows students to work on a
project in their area of interest, from conception to final prototype. Teams of three or four students are
formed around a common project idea. These teams may contain students from similar or different
concentrations, as the project allows. The Project Proposal and Feasibility Study explored the feasibility
of our chosen project, provided a structured plan for the design, and was a milestone for the project at the
end of first semester. This Final Design Report discusses the design details of the project by displaying
various decisions the team made throughout the design process. Lastly, it discusses future design work
the team would like to do in the next prototype stage.
1.1
Calvin College Engineering Department
Calvin College’s Engineering Program is ABET (Accreditation Board for Engineering and
Technology) accredited and offers a Bachelor of Science in Engineering with concentrations in Chemical,
Civil/Environmental, Electrical/Computer, and Mechanical engineering. The program also offers the
option of an international designation with any of the concentrations mentioned above. The program
“combines the foundation of a world-class liberal arts curriculum and adds the cutting-edge knowledge of
technology in engineering.” The program also “challenge[s] you to do all of [engineering] from a
reformed Christian worldview.”1
1.2
Team Member Biographies
The Eagleye team is composed of four senior engineering students in the Electrical & Computer
concentration. The team learned about a safety application involving autonomous tracking from their
faculty advisor, so they decided to pursue it due to their passion for autonomous vehicles and
quadcopters. Figure 1 shows the team picture.
Figure 1: Team Photo - Jared Zoodsma, Drew Brandsen, Robert Hoff, and Spencer Olson.
1
Below is a brief biography of each team member:
1.2.1 Jared Zoodsma
Jared hails from Grandville, Michigan where he attended Calvin Christian High School. Jared
served as the team financial manager and was in charge of researching and ordering components. In
addition to his team role, Jared primarily focused on creating a graphical user interface. Work on the GUI
included streaming video, sending GPS coordinates from the quadcopter to the Arduino, and providing a
way for the user to interact with the system. Jared enjoys working on his car, following Michigan
football, and playing sports such as soccer, ultimate, and golf. After graduation, Jared will begin his
career working at GE Aviation in Grand Rapids, Michigan.
1.2.2 Robert Hoff
Robert is from Bloomington, Illinois and served as team marketing manager. Robert was in
charge of team media including posters, website updates, and all other graphical design. Robert’s
technical design areas were the tracking algorithms and the RC control. The tracking algorithm compares
GPS locations between the vehicle and quadcopter and autonomously controls the quadcopter over
RC. Robert enjoys tinkering with electronics, learning about technology, and is an amateur photographer
in his free time. Robert has accepted a job at Northrop Grumman in Rolling Meadows, Illinois.
1.2.3 Drew Brandsen
Drew calls the great lakeside city of Holland, Michigan home. He served as the team
professional officer; assuring that all documents and materials were organized and of the utmost
quality. On the technical side, Drew was responsible for identifying and integrating quadcopter hardware
and software. Specifically, he created the software to interface between the quadcopter telemetry
information and the GUI. Drew enjoys spending time with friends, sailing, and playing intramural sports,
and after graduation will begin work with Northrop Grumman in June.
1.2.4 Spencer Olson
Acting as Project Manager, Spencer was in charge of keeping the team on track for deadlines set
by the faculty adviser, as well as setting intermediate deadlines. He was also in charge of the video
streaming and integration into the GUI, as well as the tracking algorithms and resulting flight controls
from the base station. In his free time, Spencer enjoys playing golf, camping, running, and watching
sports. Spencer will begin work this summer at Epic, in Madison, Wisconsin.
1.3
Report Format
This report details the design process of the project. Below are listed each of the chapters in the
report, along with a brief description. References are available at the end of this document.
 Chapter 2: Project Requirements - Covers the requirements of the project, both in terms of a
final product as well as the prototype that the team completed.
 Chapter 3: System Design - Evaluates different platforms for implementing a solution based on
design criteria and project requirements. This chapter shows how the team laid out their design
goals for the project.
 Chapter 4: Quadcopter Design - Discusses the quadcopter design and breaks down the
requirements, alternatives, decision criteria, and final decision for each quadcopter component;
hardware and software. This chapter also discusses implementation and tests for each
component.
2
 Chapter 5: Base Station Design - Discusses the base station design. This chapter also breaks
down each component for its requirements, alternatives, decision criteria, and final decision for
both hardware and software. Each component also has implementation and test details.
 Chapter 6: System Integration and Testing - Discusses how the various subsystems were
integrated, and how testing determined if each of those subsystems were performing their
respective tasks. The team also completed testing on the entire system to ensure its performance
met the requirements.
 Chapter 7: Project Management - This chapter explains the roles of each member of the
project. In addition, it outlines project organization, budget, and timeline.
 Chapter 8: Business Plan - The business plan discusses a potential startup company and
evaluates how this company could make a profit by selling this product.
 Chapter 9: Conclusion - The final chapter reviews the design work completed and concludes
with potential future work for the project.
1.4
Design Norms
Design norms are virtues used to shape a design process in a way that aligns with our Biblical
teachings at Calvin College. During the course of the project, the team took to heart three main design
norms that shaped the project. Below are short descriptions of each of the design norms. The team
attempted to integrate these design norms into all of our decisions. There are sometimes conflicts
attempting to satisfy these design norms, however. For example, providing a high level of justice may
result in lowered transparency. If this project was used by the United States (U.S.) military, the
information would become classified and no longer available to the public. This creates tension within
these norms that necessitated a balancing act by the team as work proceeded on the project.
1.4.1 Transparency
Transparency means that the design should be open about the communication and be
understandable throughout the design process. Transparency came into play as video-taking UAVs are
commonly highlighted in privacy concerns. The team took all actions necessary to prove that the video
taken is for responsible uses and does not show disrespect toward another party.
1.4.2 Justice
The system should be a means of achieving justice, specifically by providing additional security
for individuals who could be in life or death situations. The team made it a goal to protect those who
protect and lead us.
1.4.3 Trust
The system must be dependable in all situations. This means that within the defined
specifications, the device will function reliably and effectively
1.5
Acknowledgements
The team would like to take this opportunity to thank those who have helped make this project a
success. First, the team would like to thank DornerWorks for sponsoring the project through a donation.
This donation from DornerWorks was been a tremendous step in helping fund this project. In addition,
the team would like to thank their faculty advisor, Professor Steve VanderLeest, who was influential in
3
providing guidance and direction to the project. Next, the group would like to thank the Calvin College
Engineering program as a whole for supporting successful projects with finances and challenging students
to strive toward stronger projects. The team would also like to thank Professor Yoon Kim for allowing
them to run tests on his quadcopter and use parts of it for their project. Professor Kim also donated
several quadcopter components, which helped the team stay under budget. Justin TeBrake is another
individual who donated a net to the team and deserves gratitude. The team is also thankful to Andrew Jo
for his significant contribution to the business plan. Lastly, the team would like to thank our families for
their continued love and support in our education and personal development.
4
2
Project Requirements
This chapter identifies potential customers, defines customer requirements, and establishes a list
of deliverables for the customer. Following the requirements, the next chapter will discuss in detail the
team’s solution to solving this problem.
2.1
Problem Statement
Security is paramount for presidential motorcades, military convoys, and other safety critical
transports. Visibility is often impaired for those inside the vehicles, making it difficult to see all possible
threats ahead, behind, and to the side. Lack of visibility creates a significant danger from insurgents,
RPGs, IEDs, and other threats. Travel routes can span hundreds of miles and covering these routes with
explosive detectors, bomb-sniffing dogs, or law enforcement is costly, but still does not guarantee
complete safety. Improved visibility for individuals in the vehicles can help mitigate these external risks
2.1.1 Solution
The solution that our team proposes provides a system that shadows a motorcade or convoy from
above in order to provide a bird’s-eye view of their surroundings, enabling users to spot would-be
attackers, IEDs and other threats. Providing this real time overhead vantage point will increase the
situational awareness of those in the vehicles. This additional information provides one more level of
security that ultimately leads to justice and safety for those who need protection.
2.1.2 Target Customers
Our solution targets organizations or individuals with a need for security when traveling by land.
Our customers will include diplomats, the Secret Service and the U.S. Military to name a few. The team
will design the system to bolster security for those with this need, particularly those who serve our
country. The team considers it a just cause to protect our customer from malicious attack.
2.2
Customer Requirements
This section articulates customer-defined requirements by sub-sections such as performance,
mechanical features, and customer use. The team also identified requirements used to build prototype
1.0, which was delivered on May 10, 2014. These requirements are defined generally here, and are
applied more specifically for the chosen solution in chapters four and five.
2.2.1
Performance
These requirements define the performance of the system such as flight time and maneuverability.
2.2.1.1
Flight time
REQ 2.2.1.1.A: The airborne component of the system shall stay aloft for a minimum time of thirty
minutes.
This time is viewed as the longest journey that our customer might take where they would expect the
product to operate continuously without a recharge. Anything less and the customer would likely
consider it difficult and cumbersome to use.
REQ 2.2.1.1.A.P1 Prototype 1.0 shall stay aloft for ten minutes.
Due to commercially available batteries for this application, one can achieve this standard
time. Ten minutes also proves that with more time and money, a custom battery could be built to
increase this flight time to thirty minutes.
5
2.2.1.2
Flight Performance
REQ 2.2.1.2.A The airborne component of the system shall move from one point in three-dimensional
space to another in a straight line.
This is necessary in order to provide a high degree of maneuverability.
REQ 2.2.1.2.B The airborne component’s airspeed shall be no less than 100 km/hr (62 MPH).
100 km/hr was determined to be the maximum speed at which vehicles would move through populated,
urban areas where an attacker would be able to hide.
REQ 2.2.1.2.B.P1 Prototype 1.0 shall have a top airspeed of no less than 15 km/hr (9.3 MPH).
A speed of 15 km/hr will allow the team to produce a prototype that demonstrates the product’s
potential and provides a good platform for testing – without purchasing larger, more expensive
motors to reach higher speeds.
REQ 2.2.1.2.C The solution shall have a 2-D tracking tolerance of 1.8 meters (6 ft) from a point
horizontal from the source location to a point directly vertical of the destination location. This
requirement is applied for both a moving and stationary vehicle.
This tolerance was selected to be half the width of a standard road lane size.2 Thus, the system will stay
within the same lane as the vehicle it is following. This is required so that the system keeps a centered
view of the vehicle and its surroundings during flight.
REQ 2.2.1.2.C.P1 The solution shall have a 2-D tracking tolerance of 7.2 meters (23.6 ft) from a
point horizontal from the source location to a point directly vertical of the destination location.
This is the width of a standard two-land road.2 This was chosen for testing purposes to prove the
system can track a moving vehicle on common traffic roadways. See Figure 2 for a visual
depiction.
Figure 2: 2-D Horizontal Distance Requirement for prototype 1.0
6
REQ 2.2.1.2.D The system shall have an altitude tracking tolerance no greater than one meter (3.28 ft).
Since the camera will view from horizon to horizon (REQ 2.2.1.4.C), a one meter tolerance ensures
accurate tracking and stable video.
REQ 2.2.1.2.E The system shall fly as far as 200 meters (656 ft) from the vehicle without losing
communication.
200 meters is the common size of a city block in the United States and elsewhere. This distance would
allow the system to investigate the next intersection as the vehicle enters a new block.3
REQ 2.2.1.2.E.P1 The system shall fly as far as 100 meters (328 ft) from the vehicle without
losing communication.
100 meters (328 ft) is considered a good starting place for prototype 1.0.
2.2.1.3
Environment
REQ 2.2.1.3.A The system shall operate fully in the military temperature range of -55 to 125 degrees
Celsius (-67 to 257 degrees Fahrenheit).
Since the system is designed for military applications, the typical operating temperature for military grade
equipment is expected.
REQ 2.2.1.3.A.P1 Prototype 1.0 shall operate fully in a temperature range of 0 to 70 degrees
Celsius (32 to 158 degrees Fahrenheit).
In an effort to keep budget low, the team will use commercial grade parts. These parts are only
qualified for the above temperature range.
REQ 2.2.1.3.B Humidity: the system shall fly at all humidity levels.
The humidity and condensation levels will not ground the device.
REQ 2.2.1.3.B.P1 Humidity: the system shall fly at any temperature at which the ambient
temperature is above the dew point.
The humidity tolerated by the project will be any level at which there is no condensation. This
means that the actual relative humidity will vary based on the temperature.
REQ 2.2.1.3.C The system shall fly in dusty environments.
This is particularly important, as military uses will expose the device to dust and small debris, which will
limit visibility and be potentially destructive to system components.
REQ 2.2.1.3.C.P1 Not required for prototype 1.0.
This is similar to the humidity requirement, and though possible, is not required for the
prototype. Testing the effects of dust, even on individual components such as motors, potentially
adds costs to replace any broken components. The team is unable to afford these additional
components for testing given the budget constraints for prototype 1.0.
REQ 2.2.1.3.D The system shall fly in smoky environments.
This is particularly important, as military uses will expose the device to smoke, which will obstruct the
view of the vehicle on the ground.
7
REQ 2.2.1.3.D.P1 There is no smoke requirement for prototype 1.0
In order to view video in a smoky environment, a second infrared (IR) camera would be
necessary, which is outside the budgetary constraints of prototype 1.0.
REQ 2.2.1.3.E The system shall fly in winds up to 105 km/hr (65 MPH).
The device will be used outdoors, requiring that it work in many conditions. 105 km/hr is the lowest
bound for a tornado.4 The system is expected to be useful in less than ideal weather conditions, but flying
at tornado-like wind speeds is not expected.
REQ 2.2.1.3.E.P1 Prototype 1.0 shall withstand wind speeds up to 27 km/hr (17 MPH).
Prototype 1.0 shall be able to withstand the average wind speed at its production site: Grand
Rapids, MI.5
REQ 2.2.1.3.F The system shall have the ability to avoid obstacles within the flight path.
This is essential as power lines, trees, and other obstacles will be inherent in the environment for which
this system will be used. The system must avoid these obstacles without user intervention.
REQ 2.2.1.3.F.P1 There is no obstacle avoidance required for prototype 1.0
Due to cost and time constraints, obstacle avoidance is not explored in prototype 1.0. However,
the team understands that this is essential for this system to meet its full utilization.
2.2.1.4
Video Performance
REQ 2.2.1.4.A The system shall stream video in 1080p quality.
Since the goal of the system is to provide a video stream for protection, this video shall be high definition
in order to identify threats.
REQ 2.2.1.4.A.P1 Prototype 1.0 shall stream video at a quality of 240p.
240p quality is a good starting point quality for the video feed, which will show a proof of
concept.
REQ 2.2.1.4.B The system shall capture video at 30 frames per second.
At 30 fps, there is a 33.3 ms delay between frames. This is about 1/6 the reaction time of a typical
college-age student detecting a visual stimulus.6 As such, this is not the bottleneck in reacting to threats
and thus is sufficiently fast.
REQ 2.2.1.4.C The system shall have a 360 degree field of view in order to see threats from all directions;
forward, behind, left, and right, and down.
This is essential, as the quadcopter must be able to see enough of the area in order to see all of the
potential threats.
REQ 2.2.1.4.C.P1 There is no minimum field of view requirement for prototype 1.0.
So long as the system is able to send video, proof of concept is achieved. A more expensive
camera could be purchased with a wider camera angle to reach the final requirement.
REQ 2.2.1.4.D The system shall have a video feed with a latency of under 810 ms (from the time the
video is captured on the quadcopter, the video feed should display on the GUI within 810 ms seconds).
RPGs travel at about 300 m/s. The typical reaction time of a college-aged individual to detect a visual
stimulus is 190 ms.6 In this time, the rocket will have already traveled 57 meters. This does not factor in
8
a response by the individual or any latency in the video stream. As such, it would be nearly impossible to
protect against an RPG fired at close range.
It takes an individual with a weapon about a second to jump around a corner and level the weapon. Since
reaction time is 190 ms as mentioned above, that means that the maximum video latency is 810 ms (1000
ms jump – 190 ms reaction).
REQ 2.2.1.4.D.P1 The system shall have a video feed with a latency of under one second.
Since prototype 1.0 is a proof of concept, a video latency under one second will show the
usefulness of a video stream.
2.2.2 Physical
These requirements deal with the mechanical size and durability of the solution.
2.2.2.1
Size
REQ 2.2.2.1.A The system shall fit into the back of a vehicle, taking up no more than a total space of 1.3
x 1.3 x 0.4 meters (4.25 x 4.25 x 1.3 feet)
The device must be small enough to fit in existing military vehicles or else the system will not be
adopted. This conservative size ensures this is possible.
2.2.2.2
Weight
REQ 2.2.2.2.A The system shall weigh no more than 23 kg (51 lbs).
The device needs to be easily transported and be under the NIOSH lifting recommendation limit of 23
kg.7
2.2.2.3
Reparability
REQ 2.2.2.3.A The system shall have a modular design such that all components, with the exception of
individual components on a printed circuit board (PCB), are replaced with only the need of a screwdriver,
a needle nose pliers, and an Allen wrench.
Due to the rugged application, simple reparability is essential.
2.2.3 Customer
The following are requirements for how the system will interact with the customer.
2.2.3.1
Ease of Use
REQ 2.2.3.1.A The system as a whole shall autonomously track the intended vehicle.
Upon activating the system, it shall not require additional user interaction until the user is ready to
deactivate the system. This allows the user to focus on the results of the system (the video stream from
above) and not controlling the system. For security customers, this provides a level of justice in that this
product should never decrease the situational awareness of the user.
REQ 2.2.3.1.B Training for the device shall be minimal, requiring no more than one eight hour day.
The device shall be intuitive to use and the training shall not require an engineering or technical
background.
9
2.2.3.2
Safety
REQ 2.2.3.2.A A loss of communication shall result in the system landing and having flight temporarily
disabled.
In the event that communication is lost, the system will perform an automatic landing that will minimize
damage to the system and the chance of injury to others. It will also temporarily disable flight until the
user is able to retrieve the system.
2.2.3.3
User-System Interaction
REQ 2.2.3.3.A The base station shall provide a video feed to the user within a GUI.
These statistics will provide valuable information to the user on the flight of the system, even when they
cannot physically see the system themselves.
REQ 2.2.3.3.B The base station shall provide flight statistics including heading, altitude (in meters or in
feet, selectable by user), and speed (in km/hr or MPH, selectable by user) on the GUI.
These statistics will provide valuable information to the user on the flight of the system even when they
cannot physically see the system themselves.
REQ 2.2.3.3.B.P1 The prototype 1.0 GUI shall display video.
Since prototype 1.0 is simply a proof of concept, video feed will be the only requirement for the
GUI. Other flight statistics are deemed as stretch goals for proof of concept purposes.
REQ 2.2.3.3.C The user shall be able to override the autopilot from the GUI.
This will allow the user to fly the system outside of the normal tracking parameters, and allow uses such
as arriving at an intersection before the convoy or scanning a general area.
REQ 2.2.3.3.C.PI Prototype 1.0 shall only need the ability to toggle autopilot on and off.
Turning off autopilot will result in a deterministic action by the system, such as stationary
hovering.
REQ 2.2.3.3.D The user shall be able to enable and disable flight when the system is on the ground and
the throttle is turned down.
It is necessary for safety purposes to have a disarmed system that is still powered on. This will only be
performed once the device is no longer airborne.
REQ 2.2.3.3.E The user shall be able to power on and off the device when it is on the ground and
disarmed.
This important control must be available to the user.
10
2.3
Deliverables
The team provided a variety of deliverables throughout the project including reports, web
materials, and prototypes. Table 1 shows a list of the major deliverables with this project.
Table 1: Project Deliverables
Deliverable
Explanation
First Semester
Presentation 1
Initial presentation of the project providing an initial overview; five
minutes with one minute of question and answer (Q&A).
Presentation 2
Second presentation of the project; including a project brief, feasibility,
and current status; seven to nine minutes with one minute Q&A.
Team Website
Team website that has team overview, project details, progress, and
documentation. Resides on Calvin Engineering webpage; updated
regularly.
Team Poster
Team poster that describes the team and gives an overview of the
project; updated regularly.
PPFS
Defines the project proposal and the feasibility of the project, includes
requirements, budget, management, and status to name a few.
Prototype 0.1
Prototype completed by the end of first semester; includes a flying
quadcopter (provided by Professor Yoon Kim), video streaming over
Wi-Fi, some tracking and control research, and a basic GUI.
Second Semester
Presentation 3
The third presentation of the project.
Prototype 0.2
Prototype completed by the end of interim; includes a flying quadcopter
with the MutliWii (MW) flight controller.
Presentation 4
The final in-class presentation of the project
Design Notebooks
Prototype 1.0
Final Presentation
Final Design Report
Design notebooks that are completed individually documenting the
design process of each team member.
Final prototype of the project; includes a complete quadcopter system,
video streaming into a GUI, and accompanying base station for
tracking.
Final presentation of the project completed on Senior Design Night
Final design report of the project; includes many portions of the PPFS
as well as additional design details and other updates.
11
3
System Design
This chapter begins to evaluate different solutions to the problem outlined above. The chapter
evaluates several design alternatives and concludes with a final solution. The team determined that a
quadcopter with a base station was the best approach to solve the problem. High-level system details are
established in this section. Component-level requirements for the quadcopter and base station are
outlined in the following two chapters.
3.1
Decision Matrix
Before continuing on any further analysis of the problem, the team analyzed different solution
platforms. The team evaluated five potential solutions that are detailed in the following subsections. This section concludes with a decision matrix that summarizes our decision along with
justifications for the corresponding scores.
3.1.1 Possible Solutions
The five possible solutions are detailed below, followed by the different decision criteria used to
evaluate the solution.
3.1.1.1
Quadcopter
Quadcopters are increasing in popularity for applications ranging from search and rescue to aerial
photography. Quadcopters use four separate arms, each with a rotating propeller in order to achieve
stable flight. Recent technological developments in microprocessors have made possible the complex,
dynamic control systems necessary to balance a quadcopter. Given an easily expandable platform and a
high maneuverability in the air, quadcopters have proven effective for many common UAV applications
as is highlighted in an article in Popular Mechanics Magazine.8 An example picture of a quadcopter
outfitted with cameras is shown in Figure 3.9
Figure 3: Quadcopter Outfitted With Cameras9
3.1.1.2
RC Helicopter
RC helicopters (Figure 4) (either gas or electric powered) have been around for hobbyists for
decades. Sizes can range from hand size to the size of a car. RC helicopters give the ability to hover in
the air with relative stability (discussed later).
12
Figure 4: RC Helicopter10
3.1.1.3
RC Airplane
Like RC helicopters, RC airplanes (Figure 5) are not new to the hobby world. A downside to RC
airplanes is the fact that they cannot hover in one place.
Figure 5: RC Airplane11
3.1.1.4
Balloon / Blimp
Another solution to the problem would be to use a balloon or blimp to stream the video, an
example is seen in Figure 6. The device would be either towed by a rope or controlled by RC. It is likely
that this type of solution would have a small motor for maneuverability purposes. A factor to consider is
the size (1 m12 to 4.3 m13) of this type of device, making it a very easy target for enemies.
13
Figure 6: RC Blimp12
3.1.1.5
Panoramic Ball Camera
A new consumer product exists where a user throws a ball in the air and at the peak altitude, it
takes several pictures in each direction.14 The pictures are then loaded onto a computer for viewing and
stitched together to form a seamless overhead image of the surroundings. The panoramic ball camera is
shown in Figure 7 below. However, this solution was quickly abandoned as it became apparent this was
not practical. Some of the reasons it was deemed impractical include the fact that constant video would
not be provided, users would be exposed when the ball needed to be thrown, and throwing and retrieving
a ball from a moving vehicle is not feasible.
Figure 7: Panoramic Ball Camera14
14
3.1.2 Decision Criteria and Justification
Each of these solutions were compared against eight criteria that the team deemed important to
the problem. Table 2 displays the results of this decision matrix below. Criteria weights were chosen
between one and ten, with ten being the highest weighted. Weight values can be repeated. Scores, on the
other hand, are ranked between one and four with four being the best score. Scores are not repeated on
any category.
Table 2: Decision Matrix
Weight 1-10
Quadcopter
RC
Helicopter
RC
Airplane
Balloon /
Blimp
Battery / Fuel /
Time in Air
5
2
1
3
4
Durability /
Replacement Parts
6
4
3
2
1
Wind Resistance
7
4
3
2
1
Cost
4
1
3
2
4
Ease of Use
4
3
2
1
4
Maneuverability
9
4
3
2
1
Expandable
Platform
8
4
3
1
2
Safety in Hostile
Environment
10
4
3
2
1
Total:
186
145
99
100
Categories:
3.1.2.1
Battery / Fuel / Time in Air
How well the system stays in the air and for how long. Time in the air was a middle ground at a
five due to the customizability of each system. The team did not feel this was a major differentiating
characteristic because battery life and fuel capacity can be easily expanded in most systems.
3.1.2.2
Durability / Replacement Parts
How much abuse each system can take and how easy it is to replace broken parts. Durability and
replacement parts were moderately important at a six because the solution needs to be able to hold up
over time. Replacement parts deal with how easy it is to change parts (i.e. unplug and replace, tape, etc.).
In addition, this accounts for the variety of parts that could be replaced. For example, a system that only
has 30 parts, but only three different variety of parts, will score better than a system with 15 parts that are
all unique.
15
3.1.2.3
Wind Resistance
How well the system is able to counteract the effects of light to moderate winds, including wind
gusts. Wind resistance was rated moderately high at a seven due to the conditions in which the system
will be used. Often times there will be opposing or cross winds that must not affect the stability of the
system.
3.1.2.4
Cost
Expense to design, build, implement, and maintain. Cost was weighted at a four because when it
comes to safety, compromising is not an option. The best parts must be used in order to mitigate failure
whenever possible.
3.1.2.5
Ease of Use
Ease of use was weighted at only a four because it was assumed that the user will have at least
some training in implementation, and users will likely be knowledgeable in technical areas. In addition,
nearly all of the solutions will get easier as the amount of design and development increases.
3.1.2.6
Maneuverability
This category includes acceleration, flexibility, hovering, agility, and obstacle avoidance. This
category was weighted at nine out of ten because of the nature of the vehicle it will be following. In
many situations, the convoy to follow will start and stop quickly as well as turn corners and change
directions suddenly. To ensure the vehicle remains in the video frame the solution must be highly
maneuverable.
3.1.2.7
Expendable Platform
How easy it is to add cameras, sensors, and tracking hardware to the system. This was weighted
an eight out of ten because of the project’s need to carry multiple components such as communication
devices and cameras.
3.1.2.8
Safety
How dangerous is it for the user to interact with the system in a hostile environment. Interactions
the user could be expected to perform are as follows: launch the device, retrieve the device, and view the
video. This was weighted a ten out of ten because the primary goal of the project is to implement a
solution that leads to increased safety.
3.1.3 Justification
The following sections describe how each solution was scored for each decision criterion.
3.1.3.1
Battery / Time in Air
 Quadcopter - Average flight times for quadcopters with a 2200 mAh battery are about ten
minutes.15
 RC Helicopter - Average flight times for a RC helicopter with a 2200 mAh battery are around
seven minutes.16
 RC Airplane - Average flight times for a RC airplane with a 2200 mAh battery are around 20
minutes.17
16
 Balloon / Blimp - Balloons have a very long flight time when compared to the other solutions as
it is only limited by the time it takes to drain the battery from accessories being powered.
 Conclusion - Based on the following results, the RC helicopter came in last with the lowest
score, followed next by the quadcopter, and then the RC airplane. The best score was awarded to
the balloon/blimp for its ability to stay potentially airborne for days.
3.1.3.2
Durability / Replacement Parts
 Quadcopter - Since a quadcopter uses several of the same components (i.e. four motors, four
arms, four electronic stability controllers (ESCs) etc.) this makes it easy to swap out broken
components. Similar internal electrical components would be used for a quadcopter, RC
helicopter, and a RC airplane. Therefore, these components did not affect durability ratings.
Quadcopters received the best score.
 RC Helicopter - Similar to a quadcopter, except all components are unique. Therefore,
"replacement parts" receives a slightly lower score.
 RC Airplane - As discussed in the maneuverability section, RC airplanes have a low
maneuverability. For this reason, it is more likely that a significant crash could happen.
Therefore, it scored lower than the quadcopter and RC helicopter.
 Balloon / Blimp - Balloons received the lowest score since a small tear in the balloon would
render the device useless. Repairs to a torn balloon would not be simple. It would require
patching the hole and refilling the air. In order to refill with air, a user would be required to carry
suitable gasses in a compressed state.
 Conclusion - Looking at all of these options, the quadcopter took the highest score, followed next
by the RC helicopter. The RC airplane scored second to last with its higher probability for
crashes, followed by the balloon/blimp because of its large and vulnerable balloon.
3.1.3.3
Wind Resistance
 Quadcopter - Due to quadcopters’ robust control systems, they are able to adjust to changes in
their environment by rapidly adjusting rotor speed without needing to turn the whole system.
Quadcopter control systems are different from other air vehicles because they do not have a
forward and backward direction, allowing them to move in all directions at any point.
 RC Helicopter - A RC helicopter would not tolerate gusts as well as a quadcopter due to lower
maneuverability (See Maneuverability section below).
 RC Airplane - RC airplanes, though quite resistant to steady winds, are not able to quickly adapt
to changing winds as quadcopters can, due to the lack of dynamic balancing available.
 Balloon / Blimp - A balloon has little motor control and is completely vulnerable to the slightest
change in wind. Therefore, the blimp received the lowest score in this category.
 Conclusion - The quadcopter came in first due to its small and dense nature. Closely behind
were the RC helicopter and the RC airplane. Lastly, again because of its size and limited
maneuverability, the blimp/balloon came in last.
3.1.3.4
Cost
 Quadcopter - A stable quadcopter kit is about $600.18
 RC Helicopter - A RC helicopter with a good flight controller is around $400.19
17
 RC Airplane - A RC airplane option that includes the features needed for this project including
GPS and flight planning are roughly $600.20
 Balloon / Blimp - The cost for a blimp that would cover the specifications is about close to
$150.21
 Conclusion - The costs are lowest for the blimp due to its non-complex nature. Next is the RC
helicopter. Following that is the RC airplane, and lastly the quadcopter.
3.1.3.5
Ease of Use
 Quadcopter - Quadcopters include self-balancing software that keeps them flying without user
input. With easy to implement antennae technology, they can be controlled from inside an
armored vehicle and even programmed to fly specific routes without user input.
 RC Helicopter - RC helicopters are similar to quadcopters in that they can be self-balancing as
well as include pre-programmed routes.
 RC Airplane - While they include pre-determined route flying, it is hard to quickly change
directions in RC airplanes due to their need to maintain airspeed in order to stay aloft. Another
downside is that these require some sort of takeoff distance or a hand thrown start that exposes
the user.
 Balloon / Blimp - Once the balloon is attached or takes off, it would require little effort to land or
replace batteries. Therefore, the balloon received the best score
 Conclusion - Looking at all of the options, it was easy to see that the blimp is the easiest of the
systems to use, followed by the quadcopter. Next were the RC helicopter and the RC airplane,
which requires the most attention of the powered solutions.
3.1.3.6
Maneuverability
 Quadcopter - Since quadcopters by design do not have a front facing direction, they are capable
of moving in any direction from any orientation. This allows them to change direction and speed
very quickly. A quadcopter’s ability to pitch and roll is modeled by the difference in thrusts of
two propellers multiplied by the length of an arm, as shown in Figure 8 and the two equations
below. These two factors give the quadcopter great maneuverability.22
Figure 8: Control of a Quadcopter 22
18
𝜏𝜑 = 𝑙(𝑓2 − 𝑓4 ) 𝑎𝑛𝑑 𝜏𝜗 = 𝑙(𝑓1 − 𝑓3 )
 RC Helicopter - While RC helicopters do have a front and back direction, they are still capable
of moving in any three dimensions at a given time. Since helicopters only have two propellers
(one for thrust and one for stability). RC helicopter’s directional flight is determined by the
tilting of a swash plate, which in turn tilts the primary rotor. Therefore, the directional force
available is modeled by 𝐹 ∗ cos(𝑥) where 𝐹 is the force of the primary propeller and 𝑥 is the
angle of the swash plate from the perpendicular.23
 RC Airplane - RC airplanes are not capable of stopping, hovering, or taking off with no
speed. They are also unable to change directions quickly side to side. While the RC airplane
could go into a holding pattern, this would likely be unhelpful since constant video from the same
perspective would not be provided. A similar video perspective is desired to provide the user
with a simplistic design. If the video is constantly changing perspective, the user will need to
reorient him/herself by finding north, south, east or west every time he/she views the video. In
addition, RC airplanes were not deemed maneuverable enough to fly at low altitudes in urban
areas. Such flight patterns would almost certainly lead to a collision between the device and the
surroundings. Since RC airplanes only have one propeller, they can only travel forward. RC
airplanes cannot move side to side without moving forward and cannot move backwards at all.
 Balloon / Blimp - Balloons would likely be tethered to the vehicle, but in addition, they would
likely have small motors to allow some movement. This does not allow for great
maneuverability. Small movements could be performed, but they would be by no means quick.
 Conclusion - The quadcopter excelled in this category due to the fact that it can move in any
direction in 3-D space immediately, without turning first or building up speed. Next was the RC
Helicopter, with its only limitation being that it must turn before executing some maneuvers. The
RC airplane came in third because it requires speed to move, and any increase in altitude cannot
be done instantly but instead requires routing. Following these, the balloon scored worst of the
solutions because it is slow and its large size makes it cumbersome.
3.1.3.7
Expandable Platform
 Quadcopter - Quadcopters are designed with expansion in mind. The flat and symmetrical
frame makes it easy to incorporate more hardware onto the design. They also have self-balancing
abilities that allow for minor frame imbalance, which enables hardware to be added without the
concern for perfect balance. In addition, quadcopter arms are estimated to be 0.3 m long. With
half of the space estimated as usable for expansion (to prevent interference with propellers), the
platform size is 0.045 m2. See Figure 9 for a visualization.
19
Figure 9: Platform Area of a Quadcopter
 RC Helicopter - RC helicopters have room for cameras and GPS modules, but not much
else. Due to the hull, the only room to add additional components would be on the skids along
the bottom of the RC helicopter. The skid area is estimated to be 15.25 cm by 10.16 cm giving an
area of 0.015 m2, just 33% of the estimated area available for a quadcopter. In addition to this,
the volume available for an RC helicopter is limited by the landing gear on the bottom and the
rotor on the top, while quadcopters can outfitted with much taller components without conflicting
with the rotor and landing gear.
 RC Airplane - RC airplanes are built inside a certain predefined frame. This makes it difficult to
add additional hardware, as there often is a space issue. Unlike RC helicopters, RC airplanes do
not have skid like platforms to mount extra hardware on. Balance is also more of a concern, as
RC airplanes do not have any way of dynamically counteracting balance. This would be
problematic because symmetrical weight needs to be added to balance the RC airplane,
potentially leading to weight being added solely for balance.
 Balloon / Blimp - The balloon/blimp scored quite low in this category, not because there is not
space to add accessories, but because the lift is heavily affected by small weight increases. The
whole system may need to be enlarged in order to handle the weight of additional accessories.
 Conclusion - The quadcopter scored best in this category because it is made with expansion in
mind. The RC helicopter was second because it is similar in nature to the quadcopter, except that
the hull limits space. A RC airplane is quite space limiting because of its specifically designed
fuselage. Last are blimps, which are heavily affected by weight.
20
3.1.3.8
Safety
 Quadcopter - Quadcopters can be remotely controlled by a RC transmitter inside a
vehicle. They can be programmed to take off and land from the top of a vehicle or a platform on
the back to ensure that no operators need to exit the vehicle.
 RC Helicopter - RC helicopters have similar safety characteristics to quadcopters.
 RC Airplane - RC airplanes either require a takeoff runway, or close user interaction (for a hand
thrown start). These options extend the time required for takeoff and put the user in more danger
by requiring them to be outside the vehicle. Any time that the operator is outside of the vehicle
presents a safety risk.
 Balloon / Blimp - RC balloons or blimps would not require an operator to exit the vehicle and
expose himself or herself. However, being an air-holding vessel, these options would be
vulnerable to punctures from surrounding objects, such as trees, buildings, and bridges.
 Conclusion - Quadcopters and RC helicopters are the best in this situation due to their vertical
launches and programmable flights. However, quadcopters edged out because of increase takeoff
and landing maneuverability. The balloon/blimp scored next again because of its near-vertical
takeoffs. RC airplanes, because of their long takeoff and landing requirements, scored the lowest.
3.1.4 Best Solution
As one can see from the Table 2 above, the quadcopter scored the highest, and the team believes
that it solves the problem most efficiently due to its maneuverability, expandable platform, and safety.
3.1.4.1
Recognized Weaknesses
While the solution chosen for the problem provides many positive attributes, it is not without its
weaknesses. As is a problem with all RC vehicles, quadcopters have to constantly tradeoff between
abilities and battery life. As soon as more features are added, the weight gain and processing power
needed take a sizeable amount of battery life from the overall flight time. In addition to this problem,
quadcopters are particularly vulnerable to crash damage. This could have been a major problem,
especially during initial testing phases that will require iterative safety designs.
3.2
Overall System Block Diagram
The overall system block diagrams are in the following two sections. The first section looks at
the hardware aspects of the solution and how the two subsystems interact, while the second section
focuses on the software aspects of the solution and how the software on the two separate subsystems
communicate.
3.2.1 Hardware
The overall system hardware block diagram is seen in Figure 10 and shows the two subsystems of
the solution.
One part is the base station, which is located inside the vehicle. The base station includes two
major subsystems, the controls and communication system, and the base station computer. The controls
and communication system provides the network over which video streaming can occur, as well as the
ability to send controls from the base station computer to the quadcopter. The base station computer
provides an interface for the user to interact with the system. Refer to Chapter 5 for the details of the base
station.
21
The second part of this solution is the quadcopter. This consists of a flight controller, a video
graphics processing unit (VGPU), a camera, GPS module, motors, and a power distribution system. The
GPS module receives data from the outside world while the motors provide thrust. The flight controller
and VGPU send/receive data and controls to/from the base station. All of the interface connections are
described in Table 3 below. Refer to Chapter 4 for details of the quadcopter.
Figure 10: Overall System Hardware Block Diagram
Table 3: System Hardware Block Diagram Interface Descriptions
Signal Name
Description
Protocol
User Inputs
The user inputs include the ability to turn the quadcopter on/off,
toggle on/off the autopilot, toggle on/off the video stream, and
arm/disarm the system per REQ 2.2.3.3.C, REQ 2.2.3.3.D, REQ
2.2.3.3.E. All inputs are done through the GUI on the base station
computer, using switches on the controller, or by physically plugging
in the battery on the quadcopter.
GUI Buttons,
Controller
Switches,
Battery Plug
Power
The power for the base station is provided through the battery of the
car where the base station is mounted. The car powers both the base
station computer and the controls and communication system.
15 V Car Outlet
GPS Satellite
Signal
The controls and communication system and the GPS module on the
quadcopter receive a GPS signal from global orbiting satellites.
NMEA 018324
RC Control
Signal
A control signal to control the quadcopter is sent from the controls
and communication system to the flight controller on the quadcopter.
This signal is a frequency-hopping spread spectrum signal sent over a
2.4 GHz frequency.25
2.4 GHz
22
Telemetry
Data
The telemetry data is sent over an 802.11n Wi-Fi connection using
transmission control protocol over internet protocol (TCP/IP). This
carries the GPS coordinates to the base station computer as well as
other telemetry data such as the quadcopter altitude, heading, and RC
inputs. The GPS position is refreshed at 10 Hz and data is sent each
time new GPS data is received. See section 4.2.9.2.2 for details and
necessary bandwidth calculations.
802.11n/
TCP/IP
Video Data
The video data is sent in an H.264 compressed format over an
802.11n Wi-Fi connection. The latency is determined by REQ
2.2.1.4.D and the data is sent over UDP/IP using GStreamer. Video
data is sent on a different port than the telemetry data to prevent
interference. See section 4.2.9.2.1 for bandwidth calculations.
H.264/
GStreamer
Motor Thrust
The motors provide thrust to the outside world in order to lift the
quadcopter.
Sensor Data
The flight controller has various sensors that read data of the
environment and send it to the flight controller for analysis and flight.
N/A
Environment
Data*
*The environment data received by the flight controller sensors varies for each sensor. The barometer
measures the atmospheric pressure, the magnetometer measures the earth’s magnetic field, the
accelerometer measures the acceleration of the quadcopter, and the gyroscope measures the orientation
of the quadcopter.
3.2.2 Software
The overall software block diagram is seen in Figure 11. Once again, there are two main
subsystems, software running on the quadcopter, and software running on the base station. On the
quadcopter, the VGPU is a Raspberry Pi that is running Raspbian operating system (OS). On the OS are
the two main functional software groups, video streaming and telemetry streaming. The telemetry
streaming gets its data from the GPS module and the MW flight controller (on which the flight control
software resides). The video streaming gets data from the camera that is connected to the Raspberry
Pi. Both video and telemetry streaming interface with the base station computer over a Wi-Fi
network. For video streaming, the choice of using GStreamer over UDP/IP was made because the latency
was much lower than the RTSP stream over TCP/IP and the FFmpeg stream over UDP/IP. See section
4.2.9.2.1 for a full description of the video software decision. For telemetry streaming, due to the critical
nature of all the information arriving intact, TCP/IP was chosen as opposed to UDP/IP. This will
guarantee that the data will get from the quadcopter to the base station when a path is available. This is
critical, as the telemetry data is used in the autonomous tracking algorithms. See section 4.2.9.2.2 for a
full description of telemetry streaming. On the base station computer is a GUI running on Windows 7
OS. This GUI receives both the video stream and telemetry stream. Once the GUI receives telemetry
information, it is parsed and displayed to the screen for the user as well as sent to the RC interface
module. From there, the information is parsed, autopilot commands are calculated, and RC commands
are sent to the RC transmitter. All of the interface connections are described in Table 4 below.
23
Figure 11: Overall Software System Block Diagram
Table 4: System Software Block Diagram Interface Descriptions
Signal
Description
Type
Video
Commands
These video commands are secure shell (SSH) commands
("./gstreamer.sh") to initialize a script to setup the camera or commands
("kill $(pidof raspivid)") to terminate the camera process.
SSH
Video Data
The video data is a GStreamer stream sent from the VGPU to the base
station computer via UDP/IP as described in section 4.2.9.2.1
GStreamer
Telemetry
Command
A telemetry command is sent from the GUI via a button click. This
command initializes a python script on the Raspberry Pi. This script
gathers telemetry information from the flight controller.
Telemetry
Data
The telemetry data string can contain several instances of quadcopter
telemetry information. The quadcopter sends information via TCP/IP and
the GUI collects all information available on the TCP/IP read buffer and
stores it into the telemetry data string. This string can be several lines long.
SSH
String
See Below (
Table 5)
24
User Data
User data is the collection of flight statistics printed to the GUI. This data
serves no functional purpose, but only provides the user with up to date
information.
String
See Below (
Table 5)
RC
Commands
RC commands are voltage levels sent to the RC transmitter to be sent on to
the quadcopter as defined in section 5.2.2.
Analog
Voltage
Table 5: System Protocol Definitions
Protocol
Description
Telemetry Data The telemetry data is sent as a string. This string starts with ‘A’ to identify the
beginning of a new data packet, contains a ‘W’ to signify the end of the standard
data, and a ‘Z’ to signify the end of the packet. The standard data is the data sent to
the quadcopter and contains the fields: latitude, longitude, altitude, heading, and
timestamp. Between the ‘W’ and the ‘Z’ is various debugging data. It contains the
following data values; number of satellites, RC pitch, RC roll, RC yaw, and RC
throttle.
User Data
3.3
The user data contains information about flight statistics of which the user may be
interested. This information consists of latitude and longitude difference between
the quadcopter and base station, quadcopter altitude, quadcopter heading, amount of
time that has passed since a message was sent by the quadcopter and received by the
base station, number of quadcopter satellite GPS fixes, and quadcopter RC pitch, RC
roll, RC yaw, and RC throttle values.
Feature Matrix
During the start of the first semester, the team identified different levels of features and their
implementation importance in the system. The feature matrix is below in Table 6. The following section
goes into more detail for each of these features.
Table 6: Feature Matrix
25
3.3.1 Feature Descriptions
This section attempts to provide some clarity and definition for the features shown in the feature
matrix above.
3.3.1.1
Mechanical
Mechanical features are items that deal with the mechanical structure of the quadcopter. These
features are defined below.
 Strong Frame - The strong frame will prevent the arms from flexing and give the quadcopter
more stability. This is essential because a flimsy frame will make it difficult to control the
quadcopter as well as give the user a sense of distrust. The strong frame will also provide a
greater level of durability for crashes.
 Propeller Guards - Propeller guards will prevent the propellers from being damaged by near-by
objects. This is not essential for flight, but it is a core feature to prevent damage during testing
and to provide better safety to the users during takeoff and landing.
 Robust Landing Gear - Robust landing gear will help prevent takeoff and landing damage from
occurring. This is a basic feature since normal landing gear will perform adequately, but robust
landing gear will aid in testing the automatic takeoff and landing and increase system durability.
 Crash Protection - Adding additional protection to the quadcopter in the event of a crash would
be very useful to protect the electrical and mechanical components of the quadcopter. However,
it is not essential to creating the system.
 Reach an Air Speed of 50 km/hr - Quadcopters are known for their agility, but not their
speed. Having a quadcopter reach 50 km/hr (30 MPH) will help it keep pace with the convoy in
most situations.26 For the purposes of this project, the team will strive for 15 km/hr (9.3 MPH)
and document how achieving 50km/hr would be done given more sufficient resources.
 Rainproof - A rainproof quadcopter will be one more step in creating a robust system that can
operate in all environments. This can be done by covering the electronics in a case or using
commercial waterproofing techniques like NeverWet.27
3.3.1.2
Control
Features that affect the movement, navigation, or overall control of the quadcopter are described
here.
 Motor Control - Basic control of the motors on the quadcopter is essential to balancing and
lifting the quadcopter.
 Full Motion Control - Full motion control is the ability to move the quadcopter any 3-D
direction, which will result in a more agile system.
 Autonomous Following via GPS - Once full motion control has been established, autonomous
GPS navigation is the next step in the control process. This is a core feature to the project in
order to make the project track autonomously.
 Full control via GUI - It would be desirable to combine many of the quadcopter controls in the
GUI. Control of the quadcopter is a basic feature that would be added to the GUI.
 Altitude Control - Altitude control would give the team the ability to tell the quadcopter to fly at
a set height. This is a basic step to move towards automation.
26
 Automatic Takeoff and Landing - Automatic takeoff and landing would be a big step towards
full automation. Without this feature, a user could manually control take off and switch to
automated flight once it is in the air.
 Autonomous Following via IR - This redundant form of tracking will help the quadcopter stay
in contact with the base station in the event that the GPS signal is weak or lost.
3.3.1.3
Communication
This section describes features that could be used to provide means of communication between
the quadcopter and the base station.
 RC for Control - Since RC is the current protocol used for most quadcopters, it is essential to
have in order to achieve basic flight. In addition to basic manual control, RC could be used for
automated controls.
 Wi-Fi for Video/GPS - Wi-Fi will be the primary way for the onboard computer to send
information to the ground. A wireless network will be set up to send GPS and video signals for
processing on the control computer.
 Redundant Communication - Redundant communication protects against the event RC
communication is lost. This is the next major development step after the full system is working
with basic communication.
 Emergency Communication-Loss Landing - In the case that all communication fails with the
vehicle, a landing command will be given after a timeout expires to prevent a crash.
 Total Wi-Fi System - In this system, Wi-Fi would be used in all modes of communication
(video, GPS, and controls). This could also be used as another redundant form of control.
3.3.1.4
Power System
This section describes the components, which deal with powering the quadcopter and base
station.
 Battery - Using a battery to power the quadcopter is essential for basic flight.
 Easily Accessible - A battery will need to be replaced and recharged often. Therefore, an
accessible battery is essential.
 Base Battery Charging - Integrating the battery charger into the base station will help simplify
the number of parts in the system and increase convenience. It will also make the system more
portable if a battery is able to charge while in a moving vehicle.
 Battery Monitoring via GUI - Battery monitoring is a basic feature to prevent damaging the
battery and making it available on the GUI makes the system easier to use.
 Extra-Range Battery - Because of the high performance nature of the vehicle and all the
processes running on board, power consumption will be relatively high, yielding a need for a high
capacity battery.
 Emergency Low-Battery Landing - A command will be needed to tell the quadcopter to land if
the battery is under a certain percentage. With sensitive equipment on board, a controlled landing
is better than an uncontrolled landing.
 Laser Charging - Perhaps a science fiction feature, but this could offer nearly unlimited flight
time. Because of the difficulty, it is a dream feature.
27
3.3.1.5
Sensors
The sensor section describes various sensors that could be used to collect system information
 Hovering Sensors - Sensors on the quadcopter are essential to keep it hovering
autonomously. These include accelerometers, gyroscopes, a compass, and possibly an optical
flow camera.28
 GPS - GPS is used to monitor the location of the quadcopter and the computer on the
ground. The GPS signals from both are compared and used to calculate the necessary movement
of the quadcopter.
 Altimeter - An altimeter will be necessary to implement altitude control. Since an altimeter of
some sort is necessary for altitude control, this will need to be a core feature as well. This can be
done by using a barometer, sonar sensors, or GPS.29
 Obstacle Avoidance - Obstacle avoidance is another major design step that would be needed in
order to avoid trees, bridges, light posts, and other things that could impede the flight of the
quadcopter. This could be done by using infrared proximity sensors, or an optical flow camera.
 Speedometer - A graphical speedometer would make it possible to monitor the speed of the
quadcopter on the GUI. This is not necessary to meet the requirements, but would be useful to
users. The speed would be calculated in software using GPS data.
 Computer Vision for IR Beacon - One way of determining movement is to have an IR beacon
on the car and use computer vision to register the beacon and follow it. Similar projects have
used IR sensors from popular gaming consoles to do tracking of a vehicle.30
3.3.1.6
Video
This section defines items relating to the video, which is sent from the quadcopter to the base
station.
 Ability to Take Pictures – At a minimum, the system should provide still pictures of the area
from the view of the quadcopter.
 Stream SD (Standard Definition) Video - Core to the system is streaming video provided from
the quadcopter, which allows real-time view of the area surrounding the convoy.
 Stream Video to GUI - The streaming video should be integrated into the GUI in order to
simplify and minimize the number of components in the system.
 Stream HD (High Definition) Video - It will be important to have high quality video from the
quadcopter in order to pick out threats from above. However, it is not essential to the quadcopter
system and is therefore not an essential feature.
 Record Video while Streaming - Being able to record the video while also streaming would give
the user playback at another time. This is not crucial for the system and is therefore a stretch
feature.
 Horizon-to-Horizon Field of View - Several cameras could be used to see all around the vehicle
on the ground. Another option would be to create a simple controllable gimbal to move the
camera’s view. This component will be beyond the scope of this project but would be needed in
a production model.
28
3.4
Feature Results
Table 7 shows the results of what the team has completed or partially completed. A description
of what was accomplished for each feature is discussed below.
Table 7: Feature Matrix Implementation
3.4.1 Feature Descriptions
This section discusses what features were completed, as well as what work was done on the
partially completed features.
3.4.1.1
Mechanical
All essential and core features were implemented in this section. Crash protection, which was a
basic feature, was partially implemented and its progress is described below
 Crash Protection - In addition to propeller guards, the team added shock absorbing foam pads to
the quadcopter legs to mitigate the damage from rough landings. In addition, the team added a
protective case to the flight controller to help protect it in the event the quadcopter flips over.
The team would like to add additional protection to the quadcopter for the event of a flip in order
to fully implement this feature.
3.4.1.2
Control
Once again, the team achieved all essential and core features. One of the basic features was
implemented, but full control via GUI was not because of time constraints. The stretch goal of automatic
takeoff and landing was begun, however, and it is discussed below.
 Automatic Takeoff and Landing - For safety purposes, the team implemented a feedback free
automatic landing sequence. This autonomous landing function slowly lowers thrust until motors
are turned off. The team used a feedback free method so this function could be called when
communication is lost. In addition to the open loop landing function, the team would like to add
a take-off and another landing function with altitude feedback enabled to fully implement this
feature.
3.4.1.3
Communication
Both of the essential features were completed for communication, but one of the core features
was only partially finished. Redundant communication was begun, but not finished due to time
constraints and necessary design work for other components. However, the team did finish the stretch
29
feature of emergency communication-loss landing. The total Wi-Fi system was also under development,
and is discussed below.
 Redundant Communication - Wi-Fi is the primary medium through which data is sent; in
addition, the team began initial research on XBee as a redundant form of communication between
the base station computer and the quadcopter. The XBee modules available at Calvin were not
sufficient for low latency telemetry streaming because they only sent data at a rate of 1-2 Hz.
Therefore, research and development was paused to focus on more critical design areas. The
team would like to take the research completed and implement XBee to finish this feature.
 Total Wi-Fi System - Video and GPS information is sent from the quadcopter to the base station
via Wi-Fi; however, the team would like to send quadcopter controls over Wi-Fi as well to fully
implement this feature and give the system another layer of redundant communication.
3.4.1.4
Power System
The team finished both essential and the one core feature for the power system. They also began
work on an extra-range battery, which was one of the basic features.
 Extra-Range Battery - The team secured several batteries of different sizes that provided
different flight times. Outside of using larger batteries, the team did not experiment with multiple
batteries in parallel to achieve longer flight times.
3.4.1.5
Sensors
All of the essential, core, and basic features for sensors were completed. The team also
researched the possibility of a speedometer, and it is partially implemented as discussed below.
 Speedometer - The team began work on implementing velocity measures of the quadcopter using
GPS location information. This is not a very precise measure of speed, but does give a rough
estimate for motion. A full implementation would include air and ground speed sensors as well
as a display in the GUI.
3.4.1.6
Video
The team completed the essential, core, and basic features for video. Research was completed on
sourcing a camera with a horizon-to-horizon field of view, but it was not finished due to cost constraints.
 Horizon-to-Horizon Field of View - While there are cameras that provide a horizon-to-horizon
field of view, they are expensive. For this reason, they were not pursued any further for prototype
1.0. This is a feature, however, that would be necessary for the production model.
30
4
Quadcopter Design
This chapter discusses the quadcopter design in detail by breaking the design down by
components. Requirements, alternatives, decision criteria, decision, implementation details, and finally
unit tests are given for each quadcopter component. The next chapter will discuss the design details of
the base station. A picture of prototype 1.0 is seen in Figure 12 below.
Figure 12: Quadcopter Prototype 1.0
4.1
Quadcopter Architecture
A block diagram of the quadcopter architecture is shown in Figure 13 below. Table 8
accompanies the block diagram and lists all of the communication interfaces between the various
subsystems. The flight controller receives power through the power distribution board and interfaces with
the flight controller sensors as well as the four ESCs. The four ESCs drive each of the four motors. The
other main component of the quadcopter subsystem is the VGPU, which interfaces with the camera, the
Wi-Fi dongle, and the flight controller. A GPS module receives data from global orbiting satellites,
interacts with the VGPU through its input/output (I/O) pins, and then the telemetry information is sent
over Wi-Fi along with the video stream.
31
Figure 13: Quadcopter Hardware Block Diagram
Table 8: Quadcopter Hardware Block Diagram Interface Descriptions
Signal
Flight Controller Power
ESC Control
Description
The power distribution system on the quadcopter
provides a 5 VDC (volts direct current) power supply to
the flight controller.
The flight controller sends a pulse width modulated
signal between one millisecond and two milliseconds to
the ESC for each motor to control the movements of the
quadcopter.31
Protocol
DC
PWM
Motor Controls
The ESCs send a three phase trapezoidal wave with
varying frequencies to their respective motors to drive
them.32 The voltage ranges from 1.15 V to 1.45 V.31
Three-Phase Power
Serial COM
The VGPU and the flight controller communicate with
each other over a serial connection in order to send
telemetry data from the flight controller to the VGPU.
USB 2.0, MSP
32
VGPU Power
The power distribution system supplies a 5 VDC +/- 0.25
V power supply to the VGPU. The maximum current
draw is 700 mA giving a maximum power consumption
of 3.7 W.33
Micro USB
Video Stream
The camera sends a raw video feed to the VGPU for
processing and preparation to be sent to the base station.
H.264 video34
GPS Data
GPS data is sent to the VGPU via its digital I/O pins.
The voltage supplied to the GPS module is 5 V with a
maximum current draw of about 48 mA resulting in
power consumption of 0.24 W. Transmit and receive pins
send the NMEA (National Marine Electronics
Association) 0183 data at a 115,200 baud rate.35
NMEA 108324
A block diagram of software communication is shown in Figure 14. You can see that the
software resides on the VGPU, of which there is a video streaming block and telemetry streaming block,
as well as the GPS module and MW flight controller. The telemetry streaming interfaces with these last
two items to gather telemetry and GPS data and sends that over TCP/IP to the base station. The video
streaming software gathers video from the VGPU camera and sends that over UDP/IP to the base
station. Both video and telemetry streaming are controlled by SSH commands sent by the base
station. All of the signals are described in Table 9 below.
Figure 14: Quadcopter Software Block Diagram
33
Table 9: Quadcopter Software Block Diagram Interface Descriptions
Signal
Type
GPS Setup
GPS setup is a command sent once to set the refresh rate of the
MediaTek GPS module to 10 Hz and the baud rate to 115,200.
PMTK
NMEA 1083
String24
GPS Data
The NMEA string is sent from the MediaTek GPS module to the
Raspberry Pi at 10 Hz.
NMEA 1083
String24
The Raspberry Pi needs to gather data from the MultiWii flight
controller using the MultiWii Serial Protocol. Using this protocol
the Raspberry Pi gets the heading, altitude, and RC input values of
the quadcopter.
MSP String
The MultiWii flight controller sends the desired data back to the
Raspberry Pi using the MultiWii Serial Protocol.
MSP String
MSP Commands
MSP Data
4.2
Description
Quadcopter Components
This section discusses the design process behind quadcopter hardware and software component
choices. Figure 15 shows a picture of some of the components that are mounted at the center of the
quadcopter.
Figure 15: Central Quadcopter Components
34
4.2.1 Flight Controller
4.2.1.1
Hardware
A flight controller is used to interpret RC controls, provide telemetry data to the base station, as
well as provide dynamic control feedback to keep the quadcopter stable during flight.
4.2.1.1.1
Requirements
The flight controller shall support GPS navigation. The flight controller shall use software that
can be manipulated by the user. That is, the team shall have access to the flight controller code and
modify it per project needs. In addition, it will aid in satisfying the following customer requirements:
REQ 2.2.1.2.A, REQ 2.2.1.2.C.P1, REQ 2.2.1.2.D, REQ 2.2.1.2.E.P1, REQ 2.2.1.3.E.P1, REQ
2.2.1.4.C.P1, REQ 2.2.3.2.A, REQ 2.2.3.3.B.P1, REQ 2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ
2.2.1.3.C.P1, REQ 2.2.2.2.A, and REQ 2.2.2.3.A.
4.2.1.1.2
Alternatives
The team investigated using AeroQuad 32, ArduPilot Mega (APM) 2.6, AutoQuad 6, MultiWii
Lite, MultiWii Plus, MultiWii Pro, PX4FMU Autopilot, and UAVX-ARM32 flight controllers. The team
did not consider all flight controllers because there so many different types, so these eight were selected
with their features laid out in Table 10.
Table 10: Flight Controller Alternatives
Controller
Cost
Gyroscope
Barometer
Magnetometer
Waypoint
Capabilities
Experience
AeroQuad3236
$149.95
MPU6000
MS5611
HMC5983-TR
Add-on
None
APM 2.637
$159.99
MPU6000
MS5611
HMC5883L
Yes
Yes
AutoQuad 638
$406.25
IDG500/
ISZ500
MP3H6115A
HMC6042/
HMC1041Z
Yes
None
MultiWii Lite39
$26.49
ITG3205
None
None
None
None
MultiWii Plus40
$27.77
ITG3205
BMP085
HMC5883L
None
Yes
MultiWii Pro41
$64.99
ITG3205
BMP085
HMC5883L
Yes
None
PX4FMU
AutoPilot42
$149.00
MPU6000
MS5611
HMC5883L
Add-on
None
UAVX-ARM3243
$96.50
MPU6050
MS5611
HMC5883L
Add-on
None
4.2.1.1.3
Decision Criteria
The team evaluated the flight controllers based on their price, sensors included, waypoint
capabilities, and previous experience with the controller. Weight difference was negligible (~5 g) so
weight was not used as a decision criteria. The firmware that the flight controller runs is important, but in
35
order to test this the team would have needed to individually purchase and test all of the alternatives,
which would have surpassed our budget and time constraints.
4.2.1.1.4
Decision
The team was able to use an APM 1.4 flight controller for prototype 0.1. This is an earlier
version of the APM 2.6, and it gave the team experience using an APM flight controller. Over interim,
the team had a chance to use the MW Plus flight controller. This controller was used for prototype 0.2,
and it proved equally effective in providing a stable flight. Both flight controllers also provided similar
flight modes such as altitude-hold, heading-hold, and various stability and GPS flight modes. MW also
provided better accessibility to the source code for the flight control software, giving the team better
access to make changes to that software. Because of the experience gained with MW and APM, the team
ruled out the other flight controllers. This choice was made to mitigate start-up time if a different flight
controller was chosen. Because the MW proved as effective as the APM, but was $130 less expensive,
the team decided to go with the MW brand.
The team’s final decision was not to go with the MW Plus, however. The team needed a flight
controller that could interface with a GPS module, and the MW Plus could not do this. Because of this,
the team chose the MW Pro flight controller. This still offered the familiarity with the MW software, as
the same source code can go on either, but it offered additional serial ports that could interface with a
GPS module as well as including more memory. Though the price of the Pro was about $30 greater than
the Plus, it included a GPS module which would have cost at least that much. This made the MW Pro an
easy choice since it could interface with a GPS module and cost about the same as the MW Plus when
GPS cost was factored separately.
4.2.1.1.5
Implementation
The team purchased the MW Pro in February and installed it onto the quadcopter inside of a
plastic case connecting it to the ESCs and the VGPU as shown in Figure 13 and Figure 15 above.
4.2.1.1.6
Unit Test
While the basic flight control of the MW was tested simply by PID (proportional-integralderivative) tuning and flying the quadcopter, there were a number of sensors that needed to be tested since
they were used for the tracking algorithms. The barometer and compass on-board the flight controller
were unit tested to ensure optimal performance. These tests and are discussed below.
4.2.1.1.6.1 Barometer
The first sensor to test was the BMP085 (barometer).41 This sensor uses atmospheric pressure to
estimate the altitude of the quadcopter in meters. This sensor is needed to implement altitude-hold, a
flight mode that attempts to keep the quadcopter at the same altitude. The team tested three methods to
improve the accuracy and precision of the barometer; control (no action), adding a foam sponge cover,
and adding a mesh cover. The mesh cover proved ineffective as it made the altitude-hold worse by
inspection. The remaining two options were tested in a number of different ways. One test was flying the
quadcopter up to two meters and seeing which method provided better altitude-hold. The results are seen
in Table 11 below.
36
Table 11: Altitude-Hold Tests with Control and Foam Covers (meters)
Control
Foam
Average
0.83
1.90
Median
0.78
1.94
Standard Dev.
0.45
0.31
Standard deviation was considered the most important metric while testing. While accurate
values would be ideal, the barometer itself measures only absolute pressure, thus it was more important
that the sensor’s standard deviation was the smallest as this signified the best hold at a given altitude. As
is shown from the results, a foam cover improved the barometer in both accuracy and precision, and thus
was implemented on the flight controller. No alternative foam covers were explored due to time
constraints for the prototype. The barometer tests are seen in Appendix A.5.1.
4.2.1.1.6.2 Compass
The other sensor that underwent unit tests on the flight controller was the HMC5883L
(compass).41 In order to track the vehicle correctly, one must know where the quadcopter is pointed so
that it can turn to follow the vehicle. Because of this, unit tests were performed to ensure the accuracy
and precision of the compass. It became quickly obvious that the compass was not operating correctly
when the first flights were completed. Under motor power, the compass value would skew over 50
degrees on a 360-degree scale. After preliminary tests, and discussions with professors on campus as well
as research online, the source of the interference was found to be magnetic fields emitted from the power
distribution board. The best way to shield against this interference was simply to increase the distance
between the compass (and the flight controller on which it is mounted) and the source of
interference. Because of this change, the VGPU was mounted directly onto the quadcopter platform and
the flight controller mounted on top of that. This final layout for prototype 1.0 is seen in Figure 15 above.
Once the quadcopter was reassembled, the compass was tested again to prove the changes. The
skew resulting from motor power was under two degrees, much improved from the original case. The
quadcopter was then flown in a heading-hold mode. The purpose of this flight mode is to keep the
quadcopter’s orientation throughout a flight. With this mode activated, compass values were recorded
and the results are seen in Figure 16 below.
37
Figure 16: Compass Values with Heading-Hold Mode
As you can see from the figure above, the quadcopter was able to hold the quadcopter at a
zero-degree heading fairly well. While the PID control of the heading-hold mode does have an effect
on this test, the important thing to notice is that under power, and with varying throttle inputs, the
quadcopter held orientation and continued to correct to zero degrees. A full list of the compass tests
completed are seen in Appendix A.5.2.
4.2.1.2
Software
The flight controller software runs on the flight control board. It controls the quadcopter and
handles all of the dynamic feedback control loops.
4.2.1.2.1
Requirements
The flight controller software shall satisfy the following customer requirements: REQ 2.2.1.2.A,
REQ 2.2.1.2.D, REQ 2.2.1.2.E.P1, REQ 2.2.1.3.E.P1, REQ 2.2.3.1.B, REQ 2.2.3.2.A, and REQ
2.2.3.3.B.P1.
4.2.1.2.2
Alternatives
The software used on the flight controller is most directly related to the hardware itself. Most
flight controllers come with their own custom flight software. The MW board has two main software
options: MultiWii and MegaPirate. Both options offer similar flight modes and overall functionality
4.2.1.2.3
Decision Criteria
The highest weighted decision criterion was the experience that the team had of the
software. Having open source software was another factor, as that allows for changes for the specific
project.
38
4.2.1.2.4
Decision
The team chose to remain with the MultiWii software installed on the flight controller. This was
used on prototype 0.2, and as such, the team had experience and was more familiar with it than
MegaPirate. However, the team did upgrade from V2.2 to V2.3 when they purchased the MW Pro. V2.3
was the most recent, stable release, so it made sense to have that software package. Another nice feature
of this new version is that the barometer data calibrates to zero each time the quadcopter is powered,
making the altitude data relative to ground during flight.
4.2.1.2.5
Unit Test
There were no specific unit tests performed on the MultiWii software. Once loaded onto the
flight controller, the quadcopter was PID tuned in order to maximize stability and flight response. The
software functioned as expected and provided stable and responsive quadcopter flight.
4.2.2 GPS Module
The GPS module is responsible for obtaining quadcopter longitude and latitude data required for
tracking. The telemetry streaming software sends this data down to the base station, which compares it to
the vehicle’s position.
4.2.2.1
4.2.2.1.1
Hardware
Requirements
The GPS module shall have the accuracy necessary to track the vehicle’s position within 7.2
meters. The GPS shall refresh its location with a frequency of 10 Hz. It shall also be capable of
interfacing over digital I/O pins. In addition, the GPS module will help satisfy the following customer
requirements: REQ 2.2.1.2.C.P1, REQ 2.2.1.3.D.P1, REQ 2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ
2.2.1.3.C.P1, REQ 2.2.2.2.A, and REQ 2.2.2.3.A.
4.2.2.1.2
Alternatives
Many GPS modules use digital I/O pins to transmit and receive data. Specifically the team
considered an Adafruit GPS breakout board, which costs $39.95.38 All alternatives considered use the
NMEA 1083 protocol. The Adafruit GPS board operates at between 3.0-5.5 V and has a maximum
current draw of 25 mA giving a typical power consumption of 152 mW at 5 V. The Adafruit board
claims an accuracy of less than three meters with a maximum refresh rate of 10 Hz.44 Another GPS board
that was considered was the MTK 3329 GPS module. This module is part of a package deal that comes
with the MW Pro flight control board. It is a GPS module with maximum current draw of 48 mA, which
at 5 V results in a power consumption of 240 mW. This module also supports a refresh rate of up to 10
Hz.45
4.2.2.1.3
Decision Criteria
Cost, refresh rate, accuracy, weight, and power were all factors affecting in choosing the best
alternative.
4.2.2.1.4
Decision
The team chose the MediaTek 3329 GPS module. Because the MediaTek GPS module came
with the flight controller, it was an easy decision. Weight was not a significant issue since the difference
in weight was not significant (~1 gram). It met the required accuracy and refresh rates, and was proven to
interface with the MW Pro flight controller. The Adafruit GPS was also purchased but for the base
station instead. This decision is discussed in section 5.2.4.
39
4.2.2.1.5
Unit Test
No formal unit tests were performed. In the initial setup of the GPS module, the latitude and
longitude values were compared to that of a map and proved accurate. As seen in Chapter 6, the GPS
module was used in tandem with full communication to test the system response and positional precision.
4.2.2.2
Software
There were only minor amounts of software tweaks that occurred for the GPS module. PTK
NMEA messages are sent in the GPS setup command that increase the baud rate of the device to 115,200
and increase the refresh rate from factory default 1 Hz to maximum 10 Hz. This increase was done to
improve the latency of the feedback in the tracking algorithms.
4.2.3 Propellers
4.2.3.1
Requirements
The propellers shall be large enough to provide adequate lift for the quadcopter, but small enough
to fit on the chosen frame. Propellers are also specific to the direction of rotation, making it necessary to
match propellers with motors. The propellers used will aid in satisfying the following customer
requirements: REQ 2.2.1.1.A.P1, REQ 2.2.1.2.B.P1, REQ 2.2.1.3.E.P1, REQ 2.2.2.1.A, REQ
2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ 2.2.1.3.C.P1, REQ 2.2.2.2.A, and REQ 2.2.2.3.A.
4.2.3.2
Alternatives
Carbon fiber or plastic propellers both fit project needs. There are propellers that range in length
from 4 to 22 inches and with pitches ranging from 2 to 12 degrees.46
4.2.3.3
Decision Criteria
The choice between plastic or carbon fiber propellers depended on multiple factors. The blades
needed to be robust enough to handle moderate collisions, balanced enough to limit vibrations, and have
appropriate length and pitch values. Motors driving the propellers are rated for specific propeller sizes, so
this was taken into effect as well. Larger propellers (and those with a higher pitch) can provide more lift
because they move more air, but they also require more power. Cost was another factor, as multiple
crashes were anticipated with a new quadcopter design.
4.2.3.4
Decision
The team choose to continue using plastic propellers. Carbon fiber propellers are more expensive
and much stronger, making them even more dangerous if they were to contact an object or a person. Due
to the nature of the project with many new components of hardware and software coming together, the
team anticipated multiple crashes. Keeping this in mind, the team chose plastic propellers to help stay
under budget while providing a safer product for the user. The team specifically selected APC propellers
for their build quality. The length chosen was 25.4 cm (10 in) with a pitch of 4.5 degrees as this provided
the best balance of lift without sacrificing stability.
4.2.3.5
Unit Test
Propellers were not tested in a specific way other than during test flights of the quadcopter. The
team observed that propellers longer than 25.4 cm and with a pitch greater than 4.5 degrees made the
quadcopter too sensitive to user input. However, the larger the propeller, the more lift the quadcopter
experienced. As such, the team settled on 25.4 cm, 4.5 degree propellers because they have good lift
while leaving the quadcopter stable.
40
4.2.4 Motors
Four motors drive the propellers and provide thrust for the quadcopter.
4.2.4.1
Requirements
The motors shall be powerful enough to spin the propellers, lift the quadcopter, and move the
quadcopter at the required speed of 50 km/hr for the production model, and 15 km/hr for prototype 1.0.
In addition, the motors will help aid in satisfying the following customer requirements: REQ
2.2.1.1.A.P1, REQ 2.2.1.2.B.P1, REQ 2.2.1.3.E.P1, REQ 2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ
2.2.1.3.C.P1, REQ 2.2.2.2.A, and REQ 2.2.2.3.A.
4.2.4.2
Alternatives
On prototype 0.1, the motors were 850 Kv brushless motors with a 3.17 mm diameter shaft. The
weight of each motor was 62 grams. The motors used on prototype 0.2 were 1200 Kv (revolutions per
Minute / Volt) brushless motors with a 3 mm diameter shaft and a weight of 57.6 grams. The max current
draw is 17A.47
4.2.4.3
Decision Criteria
The motors chosen for the final design depended on weight, power (Kv), current draw, and cost.
The motors must have a maximum current draw lower than the ESC output rating. The shaft diameter is
another factor as a thicker shaft makes for a more durable motor.
4.2.4.4
Decision
Ultimately, the decision was made to use the motors from prototype 0.2. Since the frame and
power distribution system were the same between prototype 0.2 and prototype 1.0, these motors were
already proven with these components. The team was very grateful because these motors were donated to
the project by Professor Yoon Kim.
4.2.4.5
Implementation
Motors are mounted to the end of the quadcopter’s four arms as shown in Figure 12. They are
each connected to an ESC with three wires. The order of wiring simply affects the direction that the
motor turns. As such, two motors (opposite each other) are connected to spin counter-clockwise and two
connected to spin clockwise. See the quadcopter block diagram in Figure 13 and accompanying Table 8
for more information. Figure 17 below shows the direction of the motors on prototype 1.0
41
Figure 17: Prototype 1.0 Motor Direction
4.2.4.6
Unit Test
They motors were tested to lift the quadcopter and its components successfully (1.66 kg).
Above that, the motors were able to lift an additional 0.9 kg. The prototype was able to fly over 15 km/hr
meeting the prototype requirement, but the maximum speed was not tested out of concern for a crash that
could occur during this test.
4.2.5 ESCs
ESCs regulate power, and thus speed, for each motor.
4.2.5.1
Requirements
The ESC’s shall provide enough current to power the motors effectively. This means the ESC’s
must have a higher current rating than the motors they control. In addition, the ESC’s will help satisfy the
following customer requirements: REQ 2.2.1.1.A.P1, REQ 2.2.1.2.B.P1, REQ 2.2.1.3.E.P1, REQ
2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ 2.2.1.3.C.P1, REQ 2.2.2.2.A, and REQ 2.2.2.3.A.
4.2.5.2
Alternatives
ESCs can be found with different current values and are made by several manufacturers. The
team first 3DRobotics as a vendor because of their short lead-time on components. 3DRobotics only sells
one ESC, a 20A ESC with a battery eliminator circuit (BEC) output of 5 V and 2 A for $25.99.50 The
ESCs used for prototype 0.2 were Turnigy Plush 25 A ESCs with a BEC output of 5 V and 2 A, and at a
cost of $13.44, from HobbyKing.48
4.2.5.3
Decision Criteria
The major criteria for ESCs were current output, cost, compatibility with the selected motors, and
lead-time. The ESCs also provide BEC capabilities, but this is discussed in the power distribution system
in section 4.2.7. The ESCs also needed to be compatible with the voltage of the battery used in order to
power the motors. The current output was a major criterion because the ESCs needed to provide at
minimum the same current than the motor can draw (17 A for the motors selected). Cost was a major
deciding factor for both device and shipping. Where the parts ship from was another factor, as it directly
influences lead-time on parts.
42
4.2.5.4
Decision
While both ESCs provide the required current output, the team decided to use the Turnigy ESCs
because they cost half of the price of the 3DRobotics ESCs. The disadvantage to these is their lead-time
ranging from two to four weeks as they are from HobbyKing, but because of the supply from our interim
class, this was not a major concern.
4.2.5.5
Implementation
Four ESCs were purchased, with each one being connected to a specific motor. These ESCs are
powered from the power distribution board on the quadcopter and controlled from the flight controller.
4.2.5.6
Unit Test
The ESCs met the required current draw of the motors to which they were connected. Through a
simple programming setup, the ESC’s were matched with the throttle commands of the RC transmitter.
The BEC output of each ESC was also verified with a digital multimeter to ensure it met the required 5 V
output.
4.2.6 Battery
The battery provides stable voltage, high current power to all of the components on the
quadcopter.
4.2.6.1
Requirements
The battery powering the quadcopter shall provide power for all on-board sensors and computers,
as well as the ESCs and motors. The battery shall provide power for the equipment for a minimum of ten
minutes without overheating for prototype 1.0. The battery shall also have protective circuitry to prevent
overcharging and over-discharging which can cause batteries to catch fire and explode.
In addition, the battery helps satisfy the following customer requirements: REQ 2.2.1.1.A.P1,
REQ 2.2.2.1.A, REQ 2.2.3.3.E, REQ 2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ 2.2.1.3.C.P1, REQ 2.2.2.2.A,
and REQ 2.2.2.3.A.
4.2.6.2
Alternatives
Lithium ion polymer (LiPo) batteries are the standard power source for quadcopters so the team
limited their focus to them.49 Lithium ion polymer batteries have power capacities ranging from 1000
mAh to 8000 mAh.50
4.2.6.3
Decision Criteria
Cost was a portion of the decision for the battery chosen, but voltage output, discharge rate,
capacity, and weight were major criteria as well. The battery needed to operate at a voltage of 11.1 V and
have a discharge rate of at least 30 C. Batteries with larger capacities weigh more, creating a need for
more power, and eventually diminishing the advantages of having a high capacity battery. Table 12
shows the capacity, weight, cost, and charge time of several batteries that were considered. All of the
listed batteries had the required voltage and discharge rates needed. Charge time was estimated with 5 A
at 11.1 V
43
Table 12: Battery Comparisons
Capacity (mAh)
Cost
Weight (g)
Charge Time (min)
245051
$ 19.60
218
29.4
300052
$ 21.79
269
36.0
360053
$ 24.70
321
43.2
500054
$ 39.99
414
60.0
11,975*
$ 90.00*
1050*
143.7*
*Estimated battery statistics for production model
4.2.6.4
Prototype Decision
The prototype decision was to use originally use a 3600 mAh battery, which was made available
to the team by Professor Yoon Kim. This battery was chosen because it was free to use and provided a
baseline the team could run capacity tests on in order to make a more educated purchase in the future.
The team later acquired a number of other batteries from Professor Yoon Kim ranging from 2450 mAh to
5000 mAh to enable extended flight tests with only the switch of a battery. With the 5000 mAh battery,
the team was able to reach the prototype flight time requirement of ten minutes, so this battery was the
final team decision.
4.2.6.5
Production Decision
The team would likely source a custom battery that better fits within the quadcopter packaging.
Based on the described tests in the next section, average current draw for the system was 23.95 A. With a
flight time of 30 minutes, the estimated battery size is calculated below:
23.95 𝐴 ∗ (1000
𝑚𝐴
30 𝑚𝑖𝑛
) ∗ (60 𝑚𝑖𝑛/ℎ)
𝐴
= 11,975 𝑚𝐴ℎ
This is a very large battery based off the current system specifications. The average cost per mAh was
calculated to be $0.0075. Based on this, a cost of about $90 was predicted for this battery. The average
grams per mAh was found to be 0.0877 grams/mAh. With this, the weight of the battery was estimated at
1.05 kg. This data is seen in Table 12 above. Motor tests were performed with the 5000 mAh battery,
and the motors could lift an additional 0.9 kg. This new battery would add an additional 636 g (0.636 kg)
compared to the 5000 mAh battery. While this extra weight is within the lifting ability of the motors, the
team would likely need to look for power consumption savings as well as reductions in weight in order to
meet the flight time requirement of 30 minutes.
44
4.2.6.6
Unit Test
The team had access to several different battery sizes from which they could perform battery life
tests. Prototype 1.0, which included the VGPU and camera accessories, had an average current draw of
23.95 A. With the largest 5000 mAh battery, prototype 1.0 was able to meet REQ 2.2.1.1.A.P1 for flight
time. Figure 18 below shows the flight time with the five batteries available for prototype 1.0 and Table
13 lists the results including the production battery. Documentation on all the battery life tests is found in
Appendix A.1.
Figure 18: Prototype 1.0 Flight Times with various Battery Sizes
Table 13: Battery Capacities and Flight Times
Capacity (mAh)
Flight Time (min)
2450
6.5
3000
7.55
3300
8.78
5000
10.50
11,975*
30.00
*Estimated battery size for production model
The team completed calculations to determine how many batteries are needed to keep the
production system aloft with quick battery changes every 30 minutes. Based on the charge time of 143.7
minutes and a flight time of 30 minutes, the total number of batteries needed is calculated below.
45
𝑚𝑖𝑛
𝑏𝑎𝑡𝑡𝑒𝑟𝑦
𝐵𝑎𝑡𝑡𝑒𝑟𝑖𝑒𝑠 =
= 4.79 = 5 𝐵𝑎𝑡𝑡𝑒𝑟𝑖𝑒𝑠
30.0 𝑚𝑖𝑛
A total of five batteries are needed to keep the system aloft with only quick battery changes.
143.7
4.2.7 Power Distribution System
The multiple electrical components on the quadcopter are powered by a power distribution
system that is connected to the battery.
4.2.7.1
Requirements
The power distribution system shall provide adequate and stable current and voltage that is
required for all components on the quadcopter. See Table 14 for these specific requirements. In addition,
the system uses 23.95A on average as described in Appendix A.1.
Table 14: Component Power Needs
4.2.7.2
Flight Controller41
RC Receiver55
VGPU56
ESCs48
Voltage (V)
4.5-5.5
4.5-6.5
4.75 - 5.25
12
Max Current (A)
Not Available
Not Available
0.7
25
Alternatives
There were two main alternatives for this section: a custom printed circuit board or purchasing a
power distribution board.
4.2.7.3
Decision Criteria
The decision was based on cost, size, weight, and ability to scale for more components. With the
expandable nature of the project, the power distribution system needed to be able to handle extra
components necessary for the project.
4.2.7.4
Decision
The team made the decision to first evaluate whether the existing board was able to power the
VGPU along with the ESCs and flight controller. In order to add a level of redundancy to the power, the
team chose to use a HobbyWing UBEC to power the VGPU, while using the BEC of the ESCs to power
the flight controller. This later needed to be changed as three HobbyWing UBECs supplied between 5.25
and 5.3 volts to the VGPU, instead of the 5 volts specified online, making the product unusable as that
was above the maximum 5.25 volts the VGPU can accept. Due to this malfunction, the team decided to
use an extra Turnigy ESC’s BEC to power the VGPU. Because the extra ESC was able to power the
VGPU, the team chose to continue to use the existing board from prototype 0.2
4.2.7.5
Implementation
The board is mounted on the inner portion of the quadcopter and connected to all onboard
equipment to supply power. The flight controller is powered through the original four ESC BECs while
the VGPU is powered through the extra ESC added for its BEC capabilities. This reduces the load on the
main four ESCs by up to 700 mA, as that is the maximum current draw of the VGPU.
46
4.2.7.6
Unit test
After replacing the malfunctioning UBECs with a Turnigy ESC, the power distribution system
was able to power all quadcopter components successfully.
4.2.8 Frame and Landing Gear
The frame and landing gear provide a platform upon which all quadcopter components are
mounted.
4.2.8.1
Requirements
The frame shall be strong enough to carry the components necessary to accomplish the project
and be able to withstand the stress caused during landing. The weight of the quadcopter with all
components is 1.66 kg. During a severe crash, it is estimated that the quadcopter would be falling at 2
m/s (and the landing gear would provide a deceleration distance of 4 cm due to the give in the landing
v +v
gear and frame. Using ∆x = f i ∗ t (where vf = final velocity and vi = initial velocity) and solving for t
2
1
1
gives 25 s. The quadcopter changing from a velocity of 2 m/s to 0 m/s in 25 s leads to an acceleration of
−50 m/s 2. Provided this acceleration, and mass of 1.66 kg, the force that the arms would need to endure
would be 83 N due to Newton’s second law of motion: F = ma.
In addition, the frame shall help satisfy the following customer requirements: REQ 2.2.1.2.B.P1,
REQ 2.2.1.3.E.P1, REQ 2.2.2.1.A, REQ 2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ 2.2.1.3.C.P1, REQ
2.2.2.2.A, and REQ 2.2.2.3.A.
4.2.8.2
Alternatives
The team considered a designing and building a frame and landing gear in-house, or purchasing
one. Alternatives included aluminum and carbon fiber, as well a variety of other materials that were not
seriously considered due to low availability, weight, and cost. The team also considered the vendor to
purchase from, as there are many frames available.
4.2.8.3
Decision Criteria
Price, strength, weight, rigidity, and expandability were all factors that determined the chosen
frame. Expandability is defined as how well the quadcopter can expand to accommodate new
components. Rigidity was a goal, as the initial implementation of propeller guards involved arms that
extended beyond where the propellers could reach, and fishing wire was wound from each corner. This
design ensured against collisions with flat surfaces as well as corners but had a major flaw. During an
initial test, the fishing wire came loose and was caught in a propeller, bending the arm. The final decision
needed to be one that a user can trust and depend upon. This follows one of our design norms of trust.
4.2.8.4
Decision
The team decided to purchase the frame because of the lack of a mechanical engineer on the
team, but after deciding this, Professor Kim donated a frame to the project that was strong enough to
withstand the 83 N as calculated above. The propeller guards were designed in-house as the donated
frame did not include this. Prototype 0.1 helped with the decision as the team experienced problems that
poor propeller guards and landing gear can cause. During initial stages of testing, the team flew the
quadcopter for short periods in order to verify that it was stable and controllable. During these tests, the
landing gear was able to handle small falls and minor hits without damaging the quadcopter. After major
falls however, the landing gear began bending and it was obvious that the quadcopter needed a better
means of protection (Figure 19). The propeller guards implemented in prototype 0.1 were arms that
extend past where the propeller spins, ensuring safety in a collision with a flat surface. Fishing wire was
47
used, but due to the problems that it caused, it was not used in any later prototypes. The team also found
it to be uncommon for the quadcopter to hit corners during flight, eliminating the need for a propeller
protection between the ends of the arms.
Figure 19: Broken Arm
Prototype 0.2 had propeller guards similar to that of prototype 0.1 but without the fishing wire
connecting the arms (Figure 20).
Figure 20: Propeller Guards with Vibration Problems (Prototype 0.2)
These were designed specifically to keep the quadcopter from flipping if it hit a wall at an angle,
as the flat side would straighten the quadcopter. They also kept the quadcopter from landing on any of
the sensitive components if it flipped during flight. The problem with these propeller guards was the
48
vibrations they caused. The extra aluminum used to keep it from flipping vibrated enough to loosen the
motors from the frame arms. As such, the team removed the extra extensions for prototype 1.0. Figure
21 below shows the final propeller guard implementation.
Figure 21: Final Propeller Guard Implementation (Prototype 1.0)
4.2.8.5
Implementation
The frame donated by Professor Kim was used as the base for mounting the quadcopter
components. The landing gear consisted of foam attached to the bottom of the rounded feet that came
with the frame and the final propeller guards were implemented as seen in Figure 21.
4.2.8.6
Unit Test
The frame was tested in each prototype as part of the quadcopter integration tests. The frame of
prototype 0.1 did not pass the functional test as the arms were not sturdy and the propeller guards caused
problems. Prototype 0.2’s frame was sufficient, but the propeller guards caused vibrations leading to the
final implementation that has both a sturdy frame and safe propeller guards that do not cause significant
vibrations to shake the motors loose.
4.2.9 VGPU
4.2.9.1
Hardware
The VGPU is used to collect video and GPS data and sends that data from the quadcopter to the
base station.
4.2.9.1.1
Requirements
The VGPU for this prototype shall be capable of sending 240p video and GPS coordinates to a
computer. In the production device, 1080p video is necessary to provide adequate resolution for viewing
surroundings. The device should also be capable of getting GPS coordinates from satellites and sending
them to the base station. The VGPU must be able to communicate over Wi-Fi, whether through an
49
external attachment or built-in module. If an external device is used, a universal serial bus (USB) port is
necessary to accommodate this hardware.
In addition, the VGPU helps satisfy the following customer requirements: REQ 2.2.1.4.A.PI,
REQ 2.2.1.4.B, REQ 2.2.1.4.D.P1, REQ 2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ 2.2.1.3.C.P1, REQ
2.2.2.2.A, and REQ 2.2.2.3.A.
4.2.9.1.2
Alternatives
Options include development boards such as an Arduino, a BeagleBone, or a Raspberry Pi.
Development boards are ideal for this project because they limit the amount of board and integration
design needed. These three were selected for comparison due to their popularity in the electronics
industry, their relatively small sizes, and the design team’s prior experiences. None of these modules
includes Wi-Fi modules built in, due to the cost and size associated with boards in that category.
Raspberry Pi computers are available with up to 512 MB of random access memory (RAM) and
20 usable I/O ports.56 BeagleBone development boards are available with 512 MB of RAM, as well as 92
I/O pins. Arduino development boards are available with between twenty and sixty I/O pins; however,
they only provide small amounts of RAM (less than 1 MB).57 Raspberry Pi computers can run a variety
of operating systems with the most popular one being Raspbian (a Linux environment tailored toward the
Raspberry Pi). The BeagleBone Black ships with Linux kernel 3.8 but can support a number of different
operating systems that provide a versatile environment for development.58 Both the Raspberry Pi and
BeagleBone have cameras that easily interface with the boards, with the Raspberry Pi camera costing $25
In a final implementation, the team would consider a custom board as opposed to using a
development board. The team did an initial estimation of what components would be needed for a custom
board and what the layout, weight, and cost savings might be for this custom board (Figure 22). The
estimate includes only the major components such as the connectors and integrated circuits as they are the
primary factors that make up the weight and the size of the board.
Figure 22: Custom VGPU Layout
The final dimensions of the estimate seen in Figure 22 are 50 x 31.75 mm, which is a decrease of
67% from the 85.6 x 56 mm of the Raspberry Pi. By removing the Ethernet port, 3.5 mm jack, HDMI
port, RCA jack, one of the 15-pin connectors, and 22 of the 26 GPIO pins an estimate of 24 grams was
removed, cutting the weight of the VGPU in half (the Raspberry Pi weighs 46 grams). The cost saved by
removing these components was estimated to be $8.95 by finding sourcing components on DigiKey. The
cost of making 600 four-layer boards (see section 8.8) of this size is estimated to cost $1.40 per board,
50
which is just less than half the estimated cost of the full Raspberry Pi board.61 However, assuming design
time of 500 hours at $100 dollars per hour, the total design costs would be $50,000. With a cost savings
of only $10.35 per board, the total savings would be $10.35 ∗ 600 = $6,210. Despite the weight savings
of over 50% from a custom board, the additional cost of $43,000 does not justify this work. Unless a
higher volume was produced to amortize the design cost, the Raspberry Pi would be used.
4.2.9.1.3
Decision Criteria
The major criteria for choosing a device were cost, hardware capabilities, development
environment, size, and I/O ports. There was to a tradeoff between these functions as different options
were considered.
4.2.9.1.4
Decision
The team evaluated the amount of RAM used after designing the prototype and found that about
89 MB of RAM was used to stream telemetry data along with 480p video. For the production model, the
team estimates approximately double the RAM usage. This extra RAM is used for data needed for
obstacle avoidance and other expandable features.
The team chose to use a Raspberry Pi as it has a camera made for it and a team member had some
experience using one coming into the project. Arduino boards simply do not have enough RAM to
process video streaming. The BeagleBone was ruled out because of its costly camera of $88 and lack of a
second USB port. The Raspberry Pi Model B was picked over Model A because of the increased RAM
and the extra USB port. The second USB port was important because one USB port was needed for serial
communication with the flight controller while another houses the Wi-Fi dongle.
4.2.9.1.5
Implementation
The Raspberry Pi is mounted to the platform of the quadcopter in a protective plastic case and
powered by the battery through the BEC of an ESC, which provides 5 V and up to 2 A to the board. This
is sufficient, as the maximum current drawn by the Raspberry Pi is 700 mA.
4.2.9.1.6
Unit Test
The test of the Raspberry Pi was a pass-fail test to confirm that it could be powered through the
BEC and that it was able to stream both video and telemetry data to the GUI. The Raspberry Pi passed
this test under the final power distribution implementation (see Section 4.2.7 for power distribution
design and test).
51
4.2.9.2
Software
Figure 23: VGPU Software Block Diagram
Table 15: VGPU Software Block Diagram Interface Descriptions
Signal
Description
Type
Raw Video
The raw video captured by the raspivid command is forwarded to
the GStreamer pipeline for compression to send to the base
station.
Raw Video
GPS Info
The NMEA string from the GPS module is parsed and the
relevant data is sent on to the communication function where it is
joined with the Altitude, Heading, and RC input data.
String
See Below
(Table 16)
Alt, Head, RC
The telemetry data (altitude, heading, and RC input values) from
the MultiWii is converted to string type and then sent to the
communication function where it is concatenated with the GPS
data and then sent down to the base station.
String
See Below
(Table 16)
Table 16: VGPU Protocols
Protocol
GPS Info
Description
This data includes the variables of longitude, latitude, timestamp, and number of
satellites. Longitude and latitude are in the GPS form of degrees, minutes, and
seconds, and the timestamp is coordinated universal time
Alt, Head, RC Alt is the altitude of the quadcopter recorded in meters and Head is the compass
heading of the quadcopter recorded in degrees. RC consists of pitch, roll, yaw, and
throttle values that are being received by the quadcopter.
52
4.2.9.2.1
Video Streaming
The team’s final code is available for analysis and transparency on the website under code.
Transparency is especially important in the video software as many privacy concerns arise. To combat
this, the group will notify individuals if their faces appear in video recorded by the quadcopter. This way,
video taken can be used for demonstration purposes throughout the project. As this video stream is used
for security purposes in the production model, there is a tension between justice and transparency. The
customer will not likely want to broadcast the video being captured as that could lead to security risks.
The team tried to be transparent in the design of the system, but it is ultimately up to the user how
transparent they will allow the system to be.
4.2.9.2.1.1 Requirements
The video stream for prototype 1.0 shall be of the quality of at least 240p with a latency of less
than or equal to one second as defined in REQ 2.2.1.4.A.P1, and REQ 2.2.1.4.D.P1. The final
implementation shall have 1080p quality, 60 frames per second, and a latency less than or equal to 250 ms
as well as a horizon to horizon field of view as defined in REQ 2.2.1.4.A, REQ 2.2.1.4.B, REQ 2.2.1.4.C,
and REQ 2.2.1.4.D.
4.2.9.2.1.2 Alternatives
The team investigated three separate video streaming options throughout the development of the
prototype; a stream using RTSP over TCP/IP, using GStreamer over UDP/IP, and using FFmpeg over
UDP/IP. These three alternatives were chosen because of the choice of a Raspberry Pi and its camera
module as the VGPU.
There are a number of video compressions that can be used to decrease the bandwidth necessary
to send the video. H.264 is a video compression technique that reduces the size of the data sent using a
lossy compression. Without compression, it would take a bandwidth of 18 Mbps to stream 320x240
video at 30 fps and a bandwidth of 220 Mbps to stream high definition (1280x720) video at 30 fps.
320 ∗ 240 → 76,800
𝑝𝑖𝑥𝑒𝑙𝑠
𝑓𝑟𝑎𝑚𝑒
∗8
𝑏𝑖𝑡𝑠
𝑓𝑟𝑎𝑚𝑒
𝑝𝑖𝑥𝑒𝑙𝑠
𝑏𝑖𝑡𝑠
∗ 30
𝑓𝑟𝑎𝑚𝑒𝑠
𝑠𝑒𝑐𝑜𝑛𝑑
= 18 𝑀𝑏𝑝𝑠
𝑓𝑟𝑎𝑚𝑒𝑠
1280 ∗ 720 → 921,600 𝑓𝑟𝑎𝑚𝑒 ∗ 8 𝑝𝑖𝑥𝑒𝑙 ∗ 30 𝑠𝑒𝑐𝑜𝑛𝑑 = 220 𝑀𝑏𝑝𝑠
H.264 can reduce the necessary bandwidth to approximately 1.8 Mbps (320x240) and 15 Mbps
(1280x720).62
4.2.9.2.1.3 Decision Criteria
The criteria to choose a video streaming method included video quality, video latency, and ease
of implementation into the GUI. The team saw latency as the most important criterion as a low latency is
imperative for the final application of the prototype. The video quality was next most important because
it is important to have a viewable video without many glitches.
4.2.9.2.1.4 Unit Test
All three video methods use raspivid when they are running on a Raspberry Pi, a built-in video
capturing command, with parameters providing 480p video at 30 frames per second, with the exception of
GStreamer, which is captured at 60 frames per second. They all have the option to use H.264
compression. H.264 was chosen, as it is the built-in compression type on the Raspberry Pi.
The first method of video streaming to the GUI tested was using RTSP63 via TCP/IP. The RTSP
stream was sent using a VLC library64 to be implemented into the GUI using the C# VLC library. The
53
RTSP stream was able to successfully stream 480p video to the GUI, but the latency was 1.8 seconds on
average (see section A.6.1 in the Appendix).
This high latency caused the team to pursue another video streaming option. GStreamer was the
next choice because it looked as if it could provide very low latency. Unfortunately, GStreamer could not
be implemented into a C# GUI, as the C# bindings had not been released yet. GStreamer is an open
source library for streaming of multimedia.65 GStreamer had less documentation and examples than
RTSP video streaming making it harder to stream video from the Raspberry Pi. GStreamer builds a
pipeline that encodes the video and sends it via UDP/IP. The pipeline is the way GStreamer sends the
video stream in the format required, such as specifying the H.264 format, the IP address to send to, and
the port number. The base station computer can receive the pipeline by receiving the data on the
specified UDP/IP port and then decoding it. GStreamer provided a 480p video stream with a latency of
0.2 seconds. The problem with GStreamer was that it was not compatible with C# at the time it was
originally investigated because the C# bindings were not yet released. To overcome this issue, the team
used a button in the GUI to run the compiled C++ application. See section 5.2.1.2 for further details.
Again, the team looked somewhere else for video streaming with lower latency that was
compatible with the C# GUI. Using UDP/IP to send the video seemed like a lower latency option
because UDP/IP does not require acknowledge packets to be sent as TCP/IP does. The team hoped to
decrease latency by streaming video similar to the RTSP stream over TCP/IP, but instead sending it over
UDP/IP. This led to the use of FFmpeg66 and psips. FFmpeg is a command line tool that can convert
video formats, while psips is a tool that compresses the sequence parameter set (SPS) and picture
parameter set (PPS) data into an H.264 stream.67 The SPS and PPS data is needed so the receiver knows
what type of video it is receiving and therefore properly decodes it. The RTSP protocol packages and
sends the SPS and PPS data separately, so in order to stream H.264 video using FFmpeg, the SPS and
PPS data needed to be integrated into the stream; this is what psips takes care of. The video stream using
FFmpeg successfully streamed 480p video into the GUI but with similar latency to that of the RTSP
stream (1.7 seconds).
For a comparison test between the various video streaming methods, see section 6.2 and
Appendix A.6.2
4.2.9.2.1.5 Decision
Although the team would like to have the video incorporated into the GUI, it was more important
to have a lower latency video, and this caused us to choose GStreamer as the video implementation. A
separate window is therefore needed in order to view the video.
4.2.9.2.2
Telemetry Streaming
The telemetry software involves gathering sensor data from the quadcopter flight controller and
the GPS module to provide GPS positional data for the quadcopter. This data is concatenated and sent
over TCP/IP to the base station computer. It is broken up into three main functional software groups:
GPS, MultiWii Serial Protocol, and Communication.
4.2.9.2.2.1 Requirements
The telemetry software must be able to successfully obtain altitude and compass data from the
flight controller. In addition, it must gather the GPS position of the quadcopter. All of this data must
reach the base station for tracking to occur. This satisfies requirements REQ 2.2.1.2.C.P1, REQ
2.2.1.2.E.P1.
54
4.2.9.2.2.2 Operating System
There are many OSs to consider for the Raspberry Pi such as Fedora Remix, Arch Linux, and
Raspbian. Since both telemetry streaming and video streaming are happening simultaneously, an OS is
needed to ensure both processes have time on the processor. Many of these OSs are readily available and
free, so cost and availability were not major criteria. Familiarity was a factor in choosing an operating
system, as well as the popularity and thus development community surrounding a particular OS. As such,
the team chose to use Raspbian OS because of its deep support, popularity, and team familiarity.
4.2.9.2.2.3 Language
Python is one of several programming languages available on the Raspberry Pi. Other languages
include the C family, Java, and Perl. Python was selected because of its ease of development, as well as
the fact that one of the team members had prior experience with the language. Python also had the
necessary serial and socket communication libraries necessary for this project.
4.2.9.2.2.4 GPS
The GPS section of code interfaces directly with the GPS module. The first major choice was
where to interface with the GPS module. It could either be connected to the flight controller, or
connected directly onto the VGPU. The module was originally placed onto the flight controller, but the
data was no longer received in NMEA format because it passed through the MultiWii Serial Protocol
(mentioned below). As such, there was no timestamp included with the GPS data. In order to debug
some of the latency issues in later system integration testing, the timestamp of the GPS module was
necessary. Because of this need, the GPS module was moved and connected directly to the Raspberry Pi
digital I/O pins. The GPS module used the 5 V (pin 2), Ground (pin 4), UART (universal asynchronous
receive/transmit) transmit (pin 6), and the UART receive (pin 8). Figure 24 shows a layout of these
connections.
The software first ensures the GPS module is operating at a baud rate of 115,200 and a refresh
rate of 10 Hz. This refresh rate was originally at 1 Hz, but was increased to 10 Hz in order to decrease the
communication latency. After that, the software it gathers the GPS data on the serial port. It parses the
data into separate longitude, latitude, number of satellites, and timestamp variables. With these variables
updated, the communication section of the code can access them for sending to the base station.
55
Figure 24: Raspberry Pi Digital I/O Pins
4.2.9.2.2.5 MultiWii Serial Protocol
The MultiWii Serial Protocol (MSP) is a protocol used to communicate with the MW flight
controllers. It defines specific message packets that allow external devices to communicate with the flight
controller. Figure 25 shows the format of a data packet. By changing the message type, the telemetry
code can access the many different data values from the sensors on the flight controller.
Figure 25: MultiWii Serial Protocol
The MSP section of the code handles the message passing between the Raspberry Pi and the MW
flight controller. It gets the altitude data, heading data, and RC input values from the flight controller.
This data is converted from ASCII, little endian form, and 2’s compliment resulting in an integer. Once
this is complete, the variables for altitude, heading, and RC (roll, pitch, yaw, and throttle) are updated for
access from the communication section.
The Raspberry Pi was connected to the MW flight controller through a 5 V FTDI basic breakout
board and a USB cable. From this, the breakout board is connected to the serial pins on the flight
controller, which is highlighted in green in Figure 26 below.
56
Figure 26: MultiWii Pro Connection
4.2.9.2.2.6 Communication
The communication section of the code handles the interface with the base station, specifically
the GUI running on the base station computer. At the beginning of streaming, it establishes a TCP/IP
connection with the GUI. The communication sends all of the data mentioned in the sections above. This
data is sent at 10 Hz, which is limited by the GPS module. While the GPS module is the limiting factor,
this data rate is more than adequate for the tracking needs of the system.
Table 5 discusses the complete data messages that are sent to the base station.
4.2.9.2.2.7 Unit Test
There were two main unit test performed on the telemetry streaming code. The first compared
the MSP section code to that of the open-source MW GUI (used for PID tuning and changing flight
modes). This MW GUI shows the current heading and altitude data when connected to the quadcopter.
These values were compared to those of the telemetry streaming code to prove its effectiveness in
gathering this data itself. The second test verified the GPS code by comparing the longitude and latitude
values to that of a map.
4.2.10 Camera
A camera is used to record and stream video of the ground vehicle and its surroundings.
4.2.10.1 Requirements
The camera shall provide at least 240p quality for the prototype. The camera will help satisfy the
following customer requirements: REQ 2.2.1.3.D.P1, REQ 2.2.1.4.A.PI, REQ 2.2.1.4.B, REQ
2.2.1.4.C.P1, REQ 2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ 2.2.1.3.C.P1, REQ 2.2.2.2.A, and REQ
2.2.2.3.A.
4.2.10.2 Alternatives
Some possible camera solutions are the camera module for the Raspberry Pi,68 a GoPro camera69,
or a USB webcam. USB webcams come in many shapes and sizes and range in price from $4.00 to $200.
57
4.2.10.3 Decision Criteria
The team made the camera decision based on the cost of the camera, the viewing angle, the
weight, picture quality, maximum frame rate, and the ease of integrating the camera onto the quadcopter.
Ease of integration includes mounting to the quadcopter as well as compatibility with the VGPU.
4.2.10.4 Prototype Decision
The team chose the camera module for the Raspberry Pi as it is a low cost camera. At only $25,
it was much less expensive than a $200 GoPro and in the low range of USB webcam costs. It also
integrates easily into the Raspberry Pi selected as the VGPU because it was designed for the Raspberry
Pi. The field of view of this camera is only 42 degrees (see unit test below for calculation) but this is
acceptable for the prototype.
4.2.10.5 Production Decision
For the prototype, the team is using the Raspberry Pi camera but the production design might use
a different camera based on the testing results. The Raspberry Pi camera board costs $25, has a 42-degree
viewing angle, weighs 0.13 ounces, can capture video in 1080p at 60 fps. It is also designed to integrate
into the Raspberry Pi specifically. The least expensive, current GoPro camera (Hero3 White Edition)
costs $200, with a 94-degree field of view, weighs 2.6 ounces, can capture 1080p video at 30 fps, and
does not easily integrate into a Raspberry Pi. One specific USB webcam capable of providing 1080p
video (Genius WideCam F10070) costs $41, claims a 120 degree field of view, weighs 2.9 ounces,
captures 1080p video at 30 fps and has a USB connector that could connect to a Raspberry Pi but is not
designed for one. The team selected the Raspberry Pi camera, as it is only 5% of the weight of the next
lightest alternative, costs 38% less than the webcam, captures video at the required quality and frame rate,
and is the simplest to integrate into the Raspberry Pi. To meet the viewing angle requirement, multiple
cameras could be used and the image stitched together, or a wide-angle lens adapter added.
4.2.10.6 Implementation
The camera is mounted to the front of the quadcopter with an arbitrary downward viewing angle
since there is no field of view requirement for the prototype (REQ 2.2.1.4.C.P1). This provides a good
compromise between seeing ahead and seeing below. Figure 15 shows a picture of this implementation.
4.2.10.7 Unit Test
In order to test the viewing angle of the Raspberry Pi camera, the team measured the width and
height of the field of view. The camera was set up 279 cm away from the wall and the field of view was
marked. The width was 212 cm and the height was 140 cm. Solving for the angle of view yields a view
of 42 x 28 degrees.
58
5
Base Station Design
This chapter discusses the design of the base station, which controls the quadcopter and receives
the streaming video. Hardware and software design decisions are discussed for the base station portion of
the project. A picture of the final base station is seen below in Figure 27.
Figure 27: Final Base Station Implementation
5.1
Base Station Architecture
The base station hardware block diagram, seen in Figure 28 below, is composed of several main
subsystems. Table 17 lists the signals and protocols between these subsystems. A laptop (part of the base
station) is connected to an external microcontroller board via USB. Since there are no easily
implemented products to control RC vehicles with a laptop, this board passes along controls to the RC
transmitter. A GPS module interfaces with the microcontroller to indicate the position of the base
station. Lastly, a Wi-Fi router provides the network over which streaming video and telemetry is received
from the quadcopter.
59
Figure 28: Base Station Hardware Block Diagram
Table 17: Base Station Hardware Interface Descriptions
Signal
Description
Protocol
Video Data
This is the video data signal received from the quadcopter
via the wireless router. For details on this signal, see
Figure 10 and Table 3.
802.11n
Telemetry
Data_1
This is the telemetry data signal received from the
quadcopter via the wireless router. For details on this
signal, see Figure 10 and Table 3.
802.11n
Telemetry
Data_2
This is a subset of the telemetry data received by the base
station computer. Only the GPS data and the quadcopter’s
heading are sent from the base station computer to the RC
interface module, as the other telemetry data is not needed.
Micro USB 2.0
Calculated
Data
The difference in the GPS position of the quadcopter
relative to the base station is sent from the RC interface
module to the base station computer to be displayed in the
GUI.
String
See Calculated
Data in Table 19
60
USB Power
Power
The RC interface module is powered by USB from the
base station computer
A car power inverter powers both the base station
computer and Wi-Fi router.
5V
120 VAC
60 Hz
Figure 29, below, shows the software block diagram. There are two main software packages on
the base station, based on the hardware platform: the base station computer and the RC interface module.
The GUI runs on the base station computer and it contains two main software blocks: video streaming and
telemetry streaming. The video streaming block handles the UDP/IP video stream sent from the VGPU.
Telemetry streaming receives the TCP/IP data from the VGPU (on the quadcopter) and sends it on to the
RC interface module. The GUI also sends commands to control the software on the VGPU over SSH.
The RC interface module gets the telemetry data from the GUI, performs data calculations, and
determines the correct tracking response. This response is sent over RC to control the quadcopter. The
RC interface module also handles communication with its own GPS module. Table 18 discusses all of
the signals sent within this software block. Both the wireless router and the base station computer are
powered externally from a source within the car via power inverter. The base station computer powers
the RC interface module via USB connection.
Figure 29: Base Station Software Block Diagram
61
Table 18: Base Station Software Block Diagram Interfaces
Signal
Description
Type
Quadcopter Data
After the GUI receives the telemetry data, it filters this
information to send consistent and relevant data to the RC
interface module for tracking.
String
See Below
(Table 19)
Calculated Data
The RC interface module compares the quadcopter data with
the ground GPS data to compute the differences in location.
The results of this computation is sent to the GUI via the
calculated data signal.
String
See Below
(Table 19)
Table 19: Base Station Software Protocols
Protocol
Description
Quadcopter Data
Quadcopter data is a subset of the telemetry data. Quadcopter data begins with a
single ‘A’ character and ends with a ‘W’ character. Between these characters is
information about the quadcopter. This information consists of latitude, longitude,
altitude, heading, time, number of satellites, roll, pitch, yaw, and throttle for the
quadcopter.
Calculated Data
Calculated data is a necessary part of the tracking algorithm that is sent to the GUI.
Calculated data begins with a single ‘A’ character and ends with a ‘W’ character.
Information in between these is latitude difference, longitude difference, base
station latitude, base station longitude, and time difference.
5.2
Base Station Components
This section discusses the design process for the base station hardware and software components.
5.2.1 Base Station Computer
The prototype 1.0 base station computer is a laptop running Windows 7 OS. This computer runs
the GUI, which displays video as well as sends quadcopter telemetry information to the RC interface
module. Details on the process that lead up to these decisions are in the following sections.
5.2.1.1
5.2.1.1.1
Hardware
Requirements
The base station shall be easily transportable. For prototype 1.0, the computer must be rugged
enough to fall up to one meter without failures. In addition, this will help satisfy the following customer
requirements: REQ 2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ 2.2.1.3.C.P1, REQ 2.2.1.4.A.PI, REQ
2.2.1.4.B, REQ 2.2.3.3.A, REQ 2.2.3.3.B.P1, and REQ 2.2.3.3.C.PI.
5.2.1.1.2
Alternatives
The team considered using a laptop, desktop, smart phone, tablet, single board computer, or
custom computer built into the vehicle.
62
5.2.1.1.3
Decision Criteria
The criteria used to choose a computer for this project were its ability to support a GUI,
portability, cost, and relative computing power, which is needed for the graphics processing involved in a
GUI. The computer also needed to have a display large enough for the video stream to be useful.
5.2.1.1.4
Decision
The team chose to use a laptop computer owned by a team member for this project. A desktop
has almost no portability making it an easy alternative to eliminate. Buying a laptop, smartphone, or
tablet for the project would allow it to be more optimized for the system, but would drive the cost of the
project up significantly. The team chose Jared’s Gateway since he was the one primarily developing the
GUI.
5.2.1.1.5
Implementation
The base station computer resides inside the vehicle and is connected to the RC interface module.
The team used a 300 W BESTEK® Power Inverter to power the computer. The computer requires 90 W
of power and the Wi-Fi router requires 17 W of power. Therefore, the base station components use
roughly a third (107 W) of the rated power provided by the power inverter (300 W).
5.2.1.1.6
Unit Test
The GUI ran on the base station computer throughout development and performed adequately.
5.2.1.2
Software
The base station computer contains the GUI, which runs on Windows 7 OS, but is supported on
Windows XP and later. In addition, the base station computer displays the video stream from the
quadcopter as well as flight statistics.
Figure 30: Base Station Computer Software Block Diagram
63
Table 20: Base Station Computer Software Block Diagram Interfaces
Signal
Description
Type
VLC Command
See Below (
Table 21)
Play / Stop
Initializes / terminates the VLC player plugin for the GUI.
Telemetry
Data
The telemetry data string contains several instances of quadcopter
telemetry information. The quadcopter sends information via
TCP/IP and the GUI collects all information available on the
TCP/IP read buffer and stores it into the telemetry data string. This
string can be several lines long.
String
See Below
(
The filter functions parse the telemetry data and return only the
most recent telemetry instance. All other telemetry instances
contained in the telemetry data string are outdated at this point and
are discarded. Only the most recent telemetry instance is stored to
the quadcopter data string.
String
See Below
(
Quadcopter
Data
Table 21)
Table 21)
Table 21: GUI Software Protocols
Protocol
Description
Play / Stop
When the user clicks a button to toggle video, a play or stop command is sent to
the VLC player plugin to turn on and off the video respectively. These
commands are built in to the VLCPluginClass for the playlist class.
Telemetry Data
As mentioned earlier, telemetry data is a string that starts with ‘A’ to identify the
beginning of a new data packet, contains a ‘W’ to signify the end of the standard
data. This data consists of everything on the TCP/IP buffer at the time it is read.
This telemetry data contains multiple data packets separated by these ‘A’ to ‘W’
instances.
Quadcopter Data
Quadcopter Data is the string sent from the GUI to the RC interface module. It
consists of ‘A’, quadcopter latitude, quadcopter longitude, quadcopter altitude,
quadcopter orientation, quadcopter time, and ‘Z’
5.2.1.2.1
Requirements
The GUI shall display the streaming video from the quadcopter. The final production GUI shall
show various statistics from the quadcopter such as speed, altitude, and battery life (recorded as flight
time remaining). In addition, for the production prototype, the GUI shall provide an interface to disable
autopilot and control the quadcopter manually. A final view of the GUI for prototype 1.0 is seen in
Figure 31 below. In addition, the GUI will help satisfy the following customer requirements: REQ
64
2.2.1.4.A.PI, REQ 2.2.1.4.B, REQ 2.2.1.4.D.P1, REQ 2.2.3.1.A, REQ 2.2.3.1.B, REQ 2.2.3.3.A, REQ
2.2.3.3.B.P1, REQ 2.2.3.3.C.PI, REQ 2.2.3.3.D, REQ 2.2.3.3.E.
Figure 31: Final GUI
5.2.1.2.2
Operating System
5.2.1.2.2.1 Alternatives
Since the team used a laptop for the base station computer, the team considered Windows 7,
Windows XP, Ubuntu, and Fedora for the base station OS.
5.2.1.2.2.2 Decision Criteria
Since the team chose a laptop with Intel’s core i5 processor for the base station computer during
development, the OS needed to be compatible with the Intel x86 architecture. In addition, it was useful to
choose an OS for which several development environments were available.
5.2.1.2.2.3 Decision
The team decided to use Windows 7 as the OS for the base station computer. In addition, since
the GUI was written in C# (see “Language” below), the team found it preferable to develop in Visual
Studio running on Windows, since both are maintained by Microsoft and interact well together. The team
used Windows 7 since Microsoft is no longer providing support for Windows XP. Windows would also
be used in a production model.
5.2.1.2.3
Language
5.2.1.2.3.1 Alternatives
The team explored Python and C# as the languages to implement the GUI on the base station
computer. These languages were the only ones investigated due to their high-level nature and availability
of libraries to help implement the GUI. Specifically for Python, the team looked into pygame, gui2py,
PyGUI, and Tkinter modules to create the GUI. Concerning C#, the team only looked into using Visual
Studio’s default Windows Forms Application to create the GUI.
65
5.2.1.2.3.2 Decision Criteria
Languages were evaluated based on simplicity of implementation, number of available pertinent
libraries, and aesthetics.
5.2.1.2.3.3 Decision
The team decided to use C# as the language to implement the GUI. C# has several built in
wrappers to quickly create a simple and powerful GUI such as System.Windows and System.Drawing. In
addition, there are several open source libraries to help with networking, video streaming, and sending
SSH commands such as System.Net, VLCPluginClass, GMap.NET, and Tamir.SharpSsh.
5.2.1.2.3.4 Implementation
The team created a multithreaded GUI written in C#. The GUI needed to be multithreaded in
order to handle several different tasks such as listening for TCP/IP information, playing video, printing
flight statistics, and sending telemetry information via serial simultaneously. C# has several helper files
that are automatically generated. Team software is located in the Form1.cs and Form1.Designer.cs files
of the GUI project folder located here.
5.2.1.2.4
Telemetry Streaming
5.2.1.2.4.1 Alternatives
The GUI needed to receive telemetry information via TCP/IP from the quadcopter, send this
information to the RC interface module, and receive back calculated data from the RC interface module.
The GUI could use interrupts or polling to receive information from the quadcopter and RC interface
module.
5.2.1.2.4.2 Decision criteria
The team evaluated the method of gathering telemetry information based on simplicity of code
structure and the code execution time.
5.2.1.2.4.3 Decision
The team used a polling technique to gather telemetry data from the quadcopter and calculated
data from the RC interface module. Polling was deemed as a simple and efficient technique to use.
Therefore, the team decided to use a continuously polling loop to retrieve telemetry and send it to the RC
interface module.
5.2.1.2.4.4 Implementation
In the GUI, there is a listening function (listen()) that performs the telemetry streaming. First,
this function establishes a TCP/IP connection with the quadcopter. Next, it enters a loop. In this loop, the
GUI gathers all information on the TCP/IP buffer, filters this information, sends it to the RC interface
module over serial, reads all calculated data on the serial receive buffer, displays flight statistics to the
computer screen, and repeats the loop. The amount of time it takes to repeat this loop affects the
communication latencies. Details on telemetry latency requirements are found in section 6.1.3. The
filtering described earlier in Table 18 to ensure the RC interface module receives consistent data. The
team established a communication protocol to identify individual messages on a communication buffer
(TCP/IP or serial). The team sent a single ‘A’ character before a set of information and a single ‘W’ after
the message. This allowed the team to read an entire buffer and filter the message by only passing along
the most recent ‘A’ to ‘W’ instance. The user can toggle this functionality by clicking the “toggle
telemetry” button. Note: the telemetry communication is completely independent of the video streaming.
66
However, if telemetry communication is terminated, the RC interface module will recognize it is no
longer receiving telemetry information and will return manual control to the pilot. The telemetry
information must be sent to the RC interface module as quickly as possible in order to minimize latencies
and increase the tracking robustness. Therefore, it is preferable to have a dedicated thread to perform this
function. Using Visual Studio’s performance analyzer the team discovered the GUI used 50% CPU
utilization during telemetry and video streaming. This means additional processes can run on the
computer. The number of additional processes, however, depends on their CPU demand. C# has two
main libraries for multithreading: a thread class and a backgroundworker class. The team chose to use the
backgroundworker class because it is easier to interrupt the process running on the backgroundworker
thread. In addition, when the thread is interrupted, it raises a flag signaling it was successfully canceled.
Using this functionality, the team can easily stop the listening function and reset variables once the thread
stops listening. The team initialized a backgroundworker instance to take care of this function.
5.2.1.2.4.5 Unit Test
Initially, pop up boxes were used to confirm the GUI could receive information over TCP/IP.
Next, messages were sent over serial to the RC interface module from the GUI. Once these pieces were
in place, a message received over TCP/IP was sent via serial and viewed using a serial monitoring
program (PuTTY). These steps confirmed the message was successfully received and passed through the
GUI using this technique.
Once communication was confirmed, an initial communication system test seen in section A.2.1
of the Appendix showed latency once connection was established. The team discovered TCP/IP and
serial reads only consumed one line of information, but several lines existed on the respective read
buffers. This indicated that the GUI communication was the bottleneck of the system. Therefore, the
most recent telemetry information was not always being used. To fix this, the team flushed read buffers
once communication was established to remove erroneous information and the team updated read
commands to consume all information on the read buffer rather than a single line. In doing so,
communication latencies did not increase since the most recent message was passed along. The only
trade-off to this technique is that the GUI limits the telemetry information refresh frequency. However,
the GUI refreshed information multiple times a second and did not inhibit the system from meeting its
tracking accuracy requirement. Since all the information available on a buffer was consumed, the GUI
needed to filter all but one line of telemetry information (the most recent line) for the RC interface
module. These techniques ensured the RC interface module was receiving the most current information.
5.2.1.2.5
Video Streaming
5.2.1.2.5.1 Alternatives
The team decided between viewing the video through a window in the GUI and creating a
separate pop up window to view the video. In addition, the team needed to decide how the user would
start and stop the video if desired. The video could be initialized and terminated on startup and power
down or via SSH commands performed behind the scenes.
5.2.1.2.5.2 Decision criteria
The decision of where and how to playback the video was based on simplicity for user
interaction. In addition, the team had to balance the tradeoff of making things simple for the user and
giving the user full control when it comes to initializing video.
67
5.2.1.2.5.3 Decision
The team decided to integrate the video into the main GUI window for user simplicity. This
minimizes the number of windows the user must keep track of. In addition, the team decided to give the
user more control and enable the user to turn video on and off with buttons in the GUI.
5.2.1.2.5.4 Implementation
As noted above, the team decided to give the user more control and allow the user to turn the
video on and off. This was a tradeoff balancing user control with simplicity. Since the team gave the
user more control, we wanted to make this control procedure as simple as possible. Therefore, total video
control is integrated into one button - “toggle video”. This button sends an SSH command to the
quadcopter to start or stop the camera on the quadcopter, the associated processes, and the integrated
video screen in the GUI for user convenience. Upon termination, the quadcopter camera will turn off and
the video will display a black screen.
Video taking UAVs have been the point of controversy lately. Therefore, our project presents
possible concerns with privacy, so the GUI must be clear on how the video is being used. Moreover, the
team took extra care to be transparent about the intention as well as capabilities of their product.
5.2.1.2.5.5 Unit Test
Video streaming software unit tests were performed in conjunction with the system integration
test for video streaming (see section 6.2).
5.2.1.2.6
Maps
5.2.1.2.6.1 Alternatives
The team desired to implement a map into the GUI to increase the amount of information
available to the user. The team looked into several APIs (Application programming interfaces) for C# to
integrate a map into the GUI. Some of these options were SharpMap, Googlemaps, DotSpatial, and
GMap.NET.
5.2.1.2.6.2 Decision criteria
The decision of where and how to view the map information was based on simplicity for user
interaction, ability to integrate the map in the main GUI window without opening a web browser, ability
to use the map without internet access, and ability to plot points on the map via GPS location.
5.2.1.2.6.3 Decision
The team decided to use GMap.NET since it could easily integrate into the main GUI window
and could plot several points via GPS coordinates. In addition, cached map data allows users to view the
map without internet access. The only downside is that users must cache map information while internet
access is available before removing internet connection.
5.2.1.2.6.4 Implementation
The team integrated GMap.NET into the GUI and added markers for both the quadcopter and the
base station on the map. Both quadcopter and base station passed their GPS coordinates when
communicating telemetry information. The respective GPS information is passed to a function that
updates the map (UpdateMarker()). It was required to make this function a delegate method since the
thread gathering telemetry information (via listen()) did not have access to the map since it did not create
the map. Therefore, the thread that gathered telemetry information passed the GPS coordinates along to
the delegate methods to take care of updating the map asynchronously. In addition to GPS coordinates a
68
Boolean argument, isQuad, is also passed. This Boolean determined what color the marker would be.
Both quadcopter and base station used different marker colors to avoid confusion.
5.2.1.2.6.5 Unit Test
Once both quadcopter and base station GPS coordinates were reported to the map, the team
moved each subsystem to ensure the map updated appropriately.
5.2.2 RC Interface Module
An RC interface module was needed to compare the GPS data and control the quadcopter actions
through an RC transmitter. This module translates the telemetry data being sent from the base station
computer’s USB port into the RC controls used by the quadcopter. Figure 32 shows the block diagram of
the RC interface module.
Figure 32: RC Interface Module Hardware Block Diagram
Table 22: RC Interface Module Hardware Block Diagram Interfaces
Signal
Description
Analog
Controls_1
Analog controls being output by the potentiometers of the RC
transmitter. Used for manual controls.
Analog
Controls_2
Analog controls after being voltage divided by filter board. Used for
reading manual controls.
Analog
Controls_3
Analog controls sent back to the RC transmitter to control
quadcopter actions.
5 V Power
Provides power to the Adafruit GPS module from microcontroller.
GPS Setup
GPS setup is a signal sent upon initialization via digital I/O pins.
This information sets the correct settings on the Adafruit GPS
69
Type
0-5 V DC
0-2.5 V DC
0-5 V DC
5 V DC
PMTK
NMEA 0183
Commands
PWM Control
Signal
GPS Data
5.2.2.1
module such as baud rate (115,200) and refresh rate (10 Hz).
String26
A PWM signal output by the microcontroller that is later converted
to RC controls for quadcopter
0-3.3 V
PWM
The Adafruit GPS sends GPS data to the RC interface module
including longitude and latitude.
NMEA 0183
String26
Hardware
5.2.2.1.1
Microcontroller
Part of the RC interface module is a microcontroller capable of running flight software and
driving I/O pins. This microcontroller takes in various pieces of telemetry data as well as potentiometer
values from the RC transmitter and analyzes the appropriate action based on the tracking algorithm.
5.2.2.1.1.1 Requirements
The microcontroller shall send tracking controls to the RC transmitter and receive telemetry data
from the base station computer over a serial port. This portion also helps satisfy the following customer
requirements: REQ 3.2.3.2.A, REQ 3.2.3.3.C, and REQ 3.2.3.3.C.P1.
5.2.2.1.1.2 Alternatives
The team considered the Raspberry Pi Model B, Arduino Uno, Arduino Due, Arduino Mega, and
BeagleBone Black as viable options for the RC interface module microcontroller. For a final product, a
board with the exact capabilities needed would likely be custom made to minimize cost. Table 23 shows
a list of the boards the team looked at and their hardware specifications. Note: this board is required to be
compatible with an Adafruit GPS module; however, all alternatives met this requirement.
Table 23: RC Interface Module Development Board Alternatives
Raspberry Pi B56
Arduino Uno57
Arduino Due57
Arduino Mega57
BeagleBone Black58
Price
$40
$22
$40
$40
$45
I/O (Pins)
20
20
66
70
92
Adafruit
Compatible
Yes
Yes
Yes
Yes
Yes
Pin Output
50 mA at 3.3 V
or 5 V
50 mA at
3.3 V
800 mA at
3.3 V or 5 V
50 mA at 3.3 V
3.3 V, 5 V
USB Type
2 ports A
B
Micro B
B
A
5.2.2.1.1.3 Decision Criteria
Price was a factor, as well as the number and capabilities of I/O pins to ensure that all
communication could be received from the laptop and sent to the RC transmitter. This includes serial
ports and pulse width modulation (PWM) analog outputs. Although not a requirement, it would be ideal
if developers of the flight controller and the RC interface module could use the same communication
cable (USB micro Type B) for fast interchangeability with computers used for programming. The board
70
needed to have a 5 V output pin in order to power the Adafruit GPS module. This is not shown in the
above table because all the boards met this requirement. The compatibility with the selected GPS module
had to be considered, as the GPS module needed to be connected to the RC interface module. The power
consumption was not considered for this section, as all boards were reasonably small (less than 5 watts).
5.2.2.1.1.4 Prototype Decision
The team chose to use the Arduino Due as the RC interface module microcontroller. It provided
the required amount of I/O pins with room for expandability, as well as the serial ports needed for
communication with the GPS module and base station computer. It was chosen over the BeagleBone
Black because of the team’s familiarity with Arduino boards. The Arduino Due also used a USB micro
Type B for communication with the base station computer, making it a more convenient option for
development.
5.2.2.1.1.5 Production Decision
A custom board option was considered for a final product, as it could be produced for less cost
than the other options when produced in volume. By eliminating unnecessary components, the custom
board could be approximately 5.6 x 4.8 cm resulting in an area of 27 cm2. The existing Arduino has an
area of 52 cm2, making it a reduction in size of 50%. The layout with basic components and costs is
shown in Figure 33 below. With an approximate fabrication cost of $10, the total cost comes to $23,
making it the most inexpensive and spatially efficient option for very large volumes.
Roughly estimating 500 hours for design work, at $100 dollars per hour, this would amount to
design costs of $50,000. Since the team estimates only 600 sales in the first three years, design costs
would not be amortized significantly enough to warrant this custom board since the savings would only
be ($40 − $23) ∗ 600 = $10,200. The break-even cost for this custom board would have to be $50,000
in savings. Therefore, the Arduino Due would be used for full production.
Figure 33: Custom RC Interface Module Layout
71
5.2.2.1.1.6 Implementation
The RC interface module microcontroller is implemented per the RC interface module
architecture block diagram (Figure 28 and Figure 32) and the accompanying table (Table 17 and Table
22).
5.2.2.1.1.7 Unit Test
The RC interface module microcontroller was tested to see that it could perform its various tasks.
The GPS module sent serial data to the microcontroller, which parsed data and sent the results back to the
base station computer. Next, the microcontroller was connected to the filter board (see section 5.2.2.1.2)
to verify it could read multiple analog signals and write controls simultaneously to the PWM pins. This
test was run with the quadcopter to see the motors running when the RC interface module was sending
commands.
5.2.2.1.2
Filter Board
Since the chosen microcontroller development board, and all the other possible options output
PWM signals as their analog outputs, a custom board was needed to provide signals to the RC transmitter.
In addition to filtering, the signals coming from the RC transmitter range from 0-5 V, while the
microcontroller can only read and write from 0-3.3 V. These issues brought the need to step down and up
the signals before and after the microcontroller respectively. This was initially not an issue, as the first
RC transmitters used only used a 0-3.3 V scale.
5.2.2.1.2.1 Requirements
The filter board shall be capable of converting a 0-5 V analog signal from the RC transmitter to a
signal on a 0-3.3 V scale for the RC interface module microcontroller. It must also be capable of stepping
up a 0-3.3 V signal to a 0-5 V signal linearly to go the opposite direction. The solution shall be easily
powered inside the base station, requiring no more than 5 V and a small battery in series. The board shall
also convert PWM signals to pure analog signals with no delay or jitter viewable in the RC commands as
displayed in the MW GUI. The stepped down signal shall be capable of charging the analog read
capacitors in the microcontroller quickly enough to allow for the use of four channels to control the
quadcopter.
5.2.2.1.2.2 Alternatives
Alternatives for this board included using operational amplifiers for the step up and step down
functions, using a voltage divider for the step down function and an operational amplifier for the step up,
or purchasing logic converters made for similar purposes that were presumed to have a linear output. A
low pass filter with a cutoff frequency of less than 10 Hz was needed to convert a PWM signal to analog.
10 Hz was chosen because it is well below the frequency of the Arduino PWM output of 500 Hz. The
two options for the operational amplifier used to step up would be to use a more expensive operational
amplifier with a larger output swing with a given voltage supply, or place a small battery in series with
the power supply, giving a lower range operational amplifier the supply voltage needed to output a swing
of 0-5 V. The reason a separate power supply was not considered was that the filter board was to be
powered by the microcontroller, which is capable of up to 5 V. The team also decided on whether to
make the board in house or to send it out for fabrication.
5.2.2.1.2.3 Decision Criteria
The reliability of the components was the most important aspect of the filter board as any glitch
in the filter could cause damage to the RC interface microcontroller, or send a faulty command to the
quadcopter causing a crash. This means that the voltage inputs and outputs should be constant with time
72
(meaning a lowered voltage due to battery drain was undesirable), and the final software should work
without constant adjustment and rewriting of code because of this lower voltage. Power consumption and
simplicity were also strong decision points. Resistor and capacitor component values were also chosen
based on their availability.
5.2.2.1.2.4 Decision
Initially the decision was made to purchase the step up and step down circuit (BOB-12009) from
Spark Fun due to the ease of implementation and a relatively small cost ($3). The team purchased the
parts with limited information on the uses, and once implemented into the RC interface module they
proved ineffective. When the team connected the RC interface module to the MW GUI to view the RC
values being sent to the quadcopter, the response of the signal was not linear. The details of this nonlinearity were not explored, as it was obvious the part could not be used for this purpose, and the team
wanted to move forward with a solution. It was determined from this test that the logic converter would
not work for analog signals and would only work for high or low signals. While this behavior is defined
by the part name (logic level converter), the team had hoped it used linear circuitry like an operational
amplifier to do conversions. The team had to choose a new option to step up the voltage outputs. The
logic converter was sufficient, however, for both of the auxiliary channels, as they are only high or low.
Therefore, the team implemented them into the RC interface module for these channels.
The team then decided to use a voltage divider for the step down circuit in order to get a readable
voltage for the microcontroller. While this solution wastes power, the team decided to accept this loss in
order to save on design time. We chose 51 kΩ for the step down resistor values as these were commonly
available and fulfilled the requirements. The circuit for this voltage divider is shown in the filter board
schematic (Figure 34). With the chosen resistor values, the power wasted was at most 0.12 mW (by
using𝑃 = 𝑉 2 /𝑅, where R = 51 kΩ and V = 5 V). With 2500 mAh batteries, it would take 212 minutes
assuming four potentiometers at 5 V to exhaust the batteries in the transmitter.
Choosing small resistors caused the input voltages to drop since the RC transmitter
potentiometers providing the voltage were not capable of supplying enough current. If the resistor values
were chosen to be higher, it would cause a ghosting condition to occur on the analog read functions of the
microcontroller. This is caused by the input terminal not receiving enough current. Fifty-one kΩ was the
only resistor value the team had access to that showed neither signs of ghosting or current leakage.
For the step up functionality of the filter board, an LMC6484IN rail-to-rail operational amplifier
was chosen. This was chosen over another op amp, such as the LM224n, because it did not need a battery
in series with its power supply to achieve the required voltage swing. The resistors used for the amplifier
were chosen to be 1.3 kΩ and 620 Ω in order to achieve a gain of roughly three. The team chose a gain of
three to ensure that the whole range of output was utilized for the op amp (0-5 V). These resistors were
chosen because the output swing required was dependent on the resistors, and anything higher than 1.3
kΩ would decrease the output swing of the operational amplifier. By using the logic converter for the
auxiliary switches and the operational amplifier for the four directional channels the team only needed
one operational amplifier integrated circuit.
The resistor value for the low pass filter was chosen to be 620 Ω as described above. The
capacitor was chosen to be 100 µF because it showed no delay or jitter that could be viewed in the MW
GUI. This was found by testing multiple capacitors ranging from 10 µF to 5800 µF. Choosing a higher
capacitor value led to a longer time to charge the capacitor and a delay in the output. Choosing a smaller
capacitor yielded more jitter in the final signal, which would result in unstable flights. A final circuit of
the filter board is shown in Figure 34. The cutoff frequency of the low pass filter was about 3 Hz, well
below the requirement of 10 Hz.
73
Figure 34: Filter Board Circuit
Once the team decided on all the components for the circuit, the team chose to fabricate the board
in-house to minimize lead-time and cost. A diagram showing the entire process of reading controller
potentiometers from the controller to writing RC controls is shown in Figure 35 below. Figure 36 shows
the layout of the board and Figure 37 shows a photo of the actual board.
74
Figure 35: RC Interface Module Output Diagram
75
Figure 36: Filter Board Layout
Figure 37: Filter Board Photo
5.2.2.1.2.5 Implementation
The team soldered the final components to an in-house fabricated board and used them with the
microcontroller. This design allowed for pass through functions to be written where the base station is
controlling the quadcopter but is getting the controls from the RC transmitter’s potentiometer sticks. The
voltage signal from the RC transmitter is first sent through a voltage divider, and then read by the
microcontroller analog input to an integer between 0 and 900, which is the full range of the input after the
voltage divider. Since the standard command to write to a PWM port on the microcontroller uses an 8-bit
value (0-255), the float is converted to a 0-255 scale and is written back to a PWM pin. For the tracking
algorithm, the PWM output values are calculated based on the data received by the GPS module and the
base station computer. More details on this conversion from GPS coordinates to PWM are discussed in
section 5.2.2.2.
76
5.2.2.1.2.6 Unit Test
The team wrote a test program to read potentiometer values from the RC transmitter and print
those vales on the 0-900 scale as well as the 0-255 scale to a serial monitor. As potentiometers moved up
and down, or left and right, the values on both scales corresponded to those movements, validating the
design. This pass through test would later be used for manual flights, but is not part of the final
implementation of the board.
5.2.2.2
Software
Autopilot code is executed in the base station and controls the quadcopter without the need for
human interaction. A block diagram showing the flow of the RC interface module software is shown in
Figure 38 with accompanying descriptions in Table 24. For the production prototype, obstacle avoidance
needs to be included in the software, but will not be implemented or discussed further in this project.
Figure 38: RC Interface Module Software Block Diagram
77
Table 24: RC Interface Module Software Block Diagram Interface Descriptions
Signal
Position Data
Description
Parsed car GPS and quadcopter GPS data in addition to quadcopter
heading.
Type
NMEA float
+/- 3 m
Diff Data_1
Differences in latitude/longitude and orientation calculated and
output for the GUI.
Float
Diff Data_2
Differences calculated and used in the tracking algorithm to control
the quadcopter.
Float
Quad Actions
5.2.2.2.1
A set of variables (throttle, yaw, pitch, and roll) that represent an
analog write signal going to the filter board through PWM pins.
Float from
0-255
Requirements
Controls shall be given to the quadcopter in order to fly it to within 7.2 meters of the x, y position
of the vehicle and a variable altitude above the vehicle. Figure 2 in REQ 2.2.1.2.D.P1 demonstrates this
requirement. Control methods shall also allow for manual flight mode, while still allowing for
programmed automatic takeoff and landing functions. In addition, the tracking algorithm will help satisfy
the following customer requirements: REQ 2.2.1.3.F.PI, REQ 2.2.3.1.A, REQ 2.2.3.1.B, REQ 2.2.3.2.A,
REQ 2.2.3.3.C.PI, REQ 2.2.3.3.D.
5.2.2.2.2
Alternatives
The team examined two ways to provide navigation for the quadcopter autonomously: by sending
GPS coordinates to the base station or, by sending GPS coordinates to the quadcopter.
The first way was to have the quadcopter send its GPS coordinates from the flight controller or
VGPU to the base station computer. The base station would then compare its own GPS position to that of
the quadcopter and calculate the appropriate maneuvers, sent back to the quadcopter through a RC
transmitter. This technique would allow for easier manual control of the quadcopter as well as the option
of programming automatic landing and takeoff features. In this option, the flight control software is still
used for quadcopter stability; however, it is not used for tracking and positioning.
The second way of controlling is to send the vehicle’s GPS location to the quadcopter and then
allow on-board software to compute the next flight path. While this option would be a logical step in the
direction of obstacle avoidance, it would require more work to implement manual control for debugging
purposes. This method would require dynamic modification of the waypoint software on the flight
controller. Since this feature is not included in the flight control software, it must be done directly by the
team.
5.2.2.2.3
Decision Criteria
The criteria for this decision were the bandwidth necessary for each option, the effectiveness of
the product, and ability to implement manual flight controls for debugging. Calculated values for
bandwidth requirements to stream video are shown in section 4.2.9.2.1 and show 1.8 Mbps for this
project, and up to 35 Mbps for a production version (using H.264 compression). The bandwidth to send
𝑏𝑖𝑡𝑠
the telemetry data is 5.6 Kbps (70 𝑐ℎ𝑎𝑟𝑠 ∗ 8 𝑐ℎ𝑎𝑟 ∗ 10 𝐻𝑧 = 5.6 𝐾𝑏𝑝𝑠) which makes the total bandwidth
78
for this project less than 2 Mbps and 36 Mbps for a production version. The first option requires a
slightly greater amount of data being sent back and forth through Wi-Fi since it will be sending about
twice as much telemetry data; however, it is quite small (less than 0.05%) when compared to common
router bandwidths in excess of 100 Mbps, so it is not a significant criterion for this decision. Manual
flight control override is also a necessary feature.
5.2.2.2.4
Decision
The team decided to implement the option of navigation computation on the base station (thus
sending RC signals to the quadcopter). This allowed for manual flight modes and the possibility of
automatic takeoff and landing programs as well. Since the two options proposed were not mutually
exclusive, a final product could implement the other option as a source of added accuracy and
redundancy. Note: although the other option would make it easier to add obstacle avoidance, it was not
chosen because the team knew obstacle avoidance was beyond the scope of this project.
5.2.2.2.5
Implementation
The team implemented a way of tracking that used telemetry sent down from the quadcopter.
More details on the tests of this algorithm are discussed in section 6.5.
5.2.2.3
Tracking Algorithm
Once the team chose a means of directional controlling, the team chose a following algorithm.
The following algorithm is the method the code uses in order to track the vehicle.
5.2.2.3.1
Requirements
The tracking algorithm shall be capable of tracking a stationary or moving base station within 7.2
meters. In addition, the tracking algorithm will help satisfy the following customer requirements: REQ
2.2.1.3.F.PI, REQ 2.2.3.1.A, REQ 2.2.3.1.B, REQ 2.2.3.2.A, REQ 2.2.3.3.C.PI, REQ 2.2.3.3.D.
5.2.2.3.2
Alternatives
The first following method considered was to use incremental speed responses based on the
distance from the vehicle. In this method, the quadcopter would move at its fastest speed when it is
farthest away from the base station and slowest when it is closest. Figure 39 shows a representation of
these speeds. In the figure, the red area corresponds to the highest speed, while the green area represents
the slowest speed.
79
Figure 39: Distance Dependent Speed Method
The second option considered was to accelerate toward the base station, causing the quadcopter to
move fastest as it entered the 7.2 meter radius. For a stationary base station, this method requires a way
of stopping the quadcopter without it becoming unstable, but it allows for better tracking of a moving
vehicle than the first option. For a moving base station, this method would call a velocity match function
once inside the radius. Testing information for the stop function is discussed section 6.4.5.
5.2.2.3.3
Decision Criterion
The major decision criterion was the anticipated effectiveness of the tracking algorithms. Since
the team knew it would likely not be able to test both options, they chose that which showed a promise of
being effective for a stationary and moving base station.
5.2.2.3.4
Decision
For the tracking algorithm, the team decided that the quadcopter would use the acceleration
method to follow the vehicle. This gave the quadcopter the ability to catch up to a moving vehicle, as
opposed to trailing the vehicle, like the incremental speed option.
5.2.2.3.5
Implementation
Incremental development techniques were especially important in this portion of the project.
Initially an algorithm was written to only match the GPS position of the car with the quadcopter. With
this implementation, the quadcopter would always face north, pitching to manage its north/south position,
and rolling to match its east/west position. This implementation was at a disadvantage because the
quadcopter camera was often not facing the vehicle it was following. This method was chosen for the
first implementation due to the simplicity to create a basic tracking program. In order to minimize safety
concerns during testing, the GPS position was modified in the software to have the quadcopter fly to a
position ten meters north of the base station as multiple team members were standing at the base station.
See section 6.5.3 for test results of this tracking method.
The second and final implementation used the orientation data from the quadcopter flight
controller to always face the direction of travel. This is in effect the direction of the vehicle with respect
to the quadcopter. If the quadcopter would overshoot the base station, the tracking algorithm continually
turns to face the base station based on its position. This program first calculates the angle of travel based
on the latitude and longitude differences, and yaws to point in the direction of travel. A diagram of this is
80
shown in Figure 40 below. Once the angle of the quadcopter is within ten degrees of the angle of travel,
the quadcopter starts to pitch forward continually up to 10% of maximum pitch, creating a smooth
acceleration in the direction of the vehicle. Ten degrees was chosen because values smaller were tested
and resulted in a lot of stopping as the compass of the quadcopter is not precise enough to hold orientation
within anything less than ten degrees. This implementation allows for the quadcopter to always have the
vehicle in the camera’s view. See section 6.5.4 for tests involving this orientation based tracking. Once
within three meters of the base station, the quadcopter stop function is called. Three meters was chosen
for this because it was the smallest distance that the quadcopter would not actively fly away from the base
station because of the accuracies of the GPS modules.
Figure 40: Orientation Tracking
The team chose these two implementations as a set of iterative steps instead of mutually exclusive
options. Most of the functions used in the initial tracking algorithm could also be used for the final
implementation with only small modifications.
Once the following program recognizes that the quadcopter needs to move in a direction, a value
of ten percent greater than center voltage of the potentiometer is written to the pitch channel of the RC
transmitter, causing the quadcopter to fly forward. This ten percent value was chosen by test 6.4.1.5.
Since a constant pitch or roll in the quadcopter results in an acceleration, the change in tilt results in a jerk
- the derivative of acceleration. To ensure that the quadcopter does not jerk at an unstable rate, the flight
control software slowly increments the amount the quadcopter pitches forward ending in a ten percent
increase. Because the risk of crashing was constantly present, the team did not run quantitative tests on
the maximum safe jerk. The process of reaching a maximum acceleration takes five seconds, but a user
of the code has the ability to quickly change how fast this jerk is. The team chose this jerk because it
showed good results in test 0. Note: Any changes would need testing similar to those conducted for the
existing jerk.
Once the quadcopter reaches the 7.2 meter radius of the car, it then matches the velocity of the
vehicle by jerking in the same manner. If the base station is stopped or has a velocity less than 1 m/s, the
quadcopter is stopped using a recursive stop function. One meter per second was the maximum speed the
81
team saw inherent in the GPS data when the base station was stationary. The stop function reads the
amount the quadcopter is pitching forward and reverses the direction of tilt for 0.6 seconds per test 0.
By changing the acceleration toward the base station instead of moving at a constant rate, the
follow program gives the quadcopter the ability to catch up and match the velocity of the vehicle.
Increasing the speed of the quadcopter as the distance from the base station increases never allows for the
quadcopter to catch up. This is because as the quadcopter starts to speed up, it gets closer to the base
station, thus causing it to slow down again. This effect could be mitigated for some speeds based on the
speed zones used, but there would always be a point where the problem arises. The chosen algorithm
moves the quadcopter faster as the time it has been outside of the 7.2 meter radius increases.
5.2.2.3.6
Unit Test
In order to verify the software was working, various steps were transmitted through the serial port
showing calculations and actions. Initial tests were run when the follow program was given artificial GPS
data and the differences in position were printed through the serial port. Next, the RC values were printed
at every iteration of the software loop and artificial GPS data was given in each direction, giving the
ability to check the program’s actions for any possibility of flight positions.
Since testing of the effectiveness of the tracking algorithm in a real situation requires the use of
the whole system, further tests are discussed in Chapter 6.
5.2.3 RC Transmitter
The RC transmitter is used to send controls to the flight controller on the quadcopter. It is also
used as a means for manual flight.
5.2.3.1
Hardware
5.2.3.1.1
Requirements
The RC transmitter shall be compatible with the RC receiver on the quadcopter or come with a
receiver that can be easily implemented on the quadcopter. It shall have the correct control features of
throttle, pitch, yaw, and roll to control the quadcopter. In addition, the RC transmitter will help satisfy the
following customer requirements: REQ 2.2.1.2.E.P1, REQ 2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, and REQ
2.2.1.3.C.P1.
5.2.3.1.2
Alternatives
An RC transmitter module and an RC receiver are used to send RC communication to the
quadcopter. The first option explored and tested was the Turnigy 6x transmitter. The second option
considered was the FlySky FS-CT6B.
5.2.3.1.3
Decision Criteria
Cost, lead-time, and ease of modification were considered in deciding which RC transmitter to
use.
5.2.3.1.4
Decision
A Turnigy 6X transmitter was initially used due to its immediate availability to the team.
However, after two instances of unsolvable defects, the team chose to reevaluate that decision. The
Turnigy 6X transmitter costs $30 and has a lead-time of about two to four weeks while the FlySky
transmitter costs $40 but has a lead-time of five to seven days. The lead-time was more important than
cost at the time of this decision, as well as the fear of another Turnigy transmitter being defective.
Because of this, the team chose the FlySky transmitter.
82
5.2.3.1.5
Implementation
The team incorporated the new transmitter’s receiver into the quadcopter and the transmitter was
connected to the RC interface module (see section 5.1 for details).
5.2.3.1.6
Unit Test
The team tested the RC transmitter by viewing the RC values received by the quadcopter using
the MultiWii GUI.
5.2.4 GPS Module
The base station GPS module is used to determine the geographical location of the base station.
5.2.4.1
5.2.4.1.1
Hardware
Requirements
The GPS module shall have the accuracy necessary to track the vehicle’s position within three
meters. In addition, the GPS module will help satisfy the following customer requirements: REQ
2.2.1.2.C.P1, REQ 2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ 2.2.1.3.C.P1, and REQ 2.2.1.3.D.P1.
5.2.4.1.2
Alternatives
GPS modules come with either a USB interface, or digital I/O pins to transmit and receive data.
Specifically the team considered a GlobalSat USB GPS receiver, which costs $27.89,71 and an Adafruit
GPS breakout board, which costs $39.95.44 All alternatives considered use the NMEA 1083 protocol. The
Adafruit GPS board operates between 3.0-5.5 V and has a maximum current draw of 25 mA giving a
typical power consumption of 152 mW at 5 V. The Adafruit board claims an accuracy of less than three
meters with a maximum refresh rate of 10 Hz. The GlobalSat GPS module operates between 4.5-6.5 V
with a typical current draw of 80 mA giving a typical power consumption of 400 mW.72 The GlobalSat
module claims an accuracy of about five meters and does not provide refresh rate information in the data
sheet.
5.2.4.1.3
Decision Criteria
Cost, time required for implementation, accuracy, and power are factors affecting the alternative
chosen. Time required for implementation was determined by support forums and documented projects
detailed online.
5.2.4.1.4
Decision
Because the GPS calculations are being performed on the microcontroller, it seemed intuitive to
connect the GPS directly to the microcontroller instead of receiving it on the computer through USB and
then sending it to the microcontroller. The team chose to use the Adafruit GPS board as it interfaces with
the RC interface module microcontroller selected via analog input pins, and claimed a better accuracy
than the GlobalSat GPS. In addition, the increased refresh rate of the Adafruit GPS decreased the time it
took the RC interface module to receive the GPS data and take action.
5.2.4.1.5
Implementation
The GPS module is mounted in the RC interface module and connects to the microcontroller’s
digital I/O pins to send data. Figure 41 and Figure 42 show the inside and outside of the RC interface
module, which houses the GPS module.
83
Figure 41: RC Interface Module External Photo
Figure 42: RC Interface Module Internals
5.2.4.1.6
Unit Test
The team gathered the GPS coordinates from the GPS module by printing the data on a serial
monitor and verified those with coordinates online to confirm that they are received correctly. As seen in
Chapter 6, the GPS module was used in tandem with full communication to test the system response and
positional precision.
5.2.4.2
Software
There were only minor amounts of software changes needed for the GPS module. The RC
interface module microcontroller sends PMTK NMEA messages in the GPS setup command that
increases the baud rate of the device from 9600 to 115,200 and increases the refresh rate from the factory
84
default 1 Hz to a maximum 10 Hz. The motivation for this change was to improve the latency of the
feedback in the tracking algorithms (See section 6.1.3).
5.2.5 Wireless Router
The wireless router receives the video and telemetry information from the quadcopter.
5.2.5.1
5.2.5.1.1
Hardware
Requirements
The wireless router used in the system shall be capable of sending multiple types of information
over simultaneous wireless channels. This includes sending video, quadcopter telemetry information, and
controls from the base station. The router must provide the necessary bandwidth to stream HD video and
telemetry data. To send 240p video for the prototype the router must have a bandwidth of 1.8 Mbps (see
section 4.2.9.2.1 for calculations). Continually sending GPS data takes a bandwidth of 5.6 Kbps (see
section 5.2.2.2.3 for this calculation). This gives a total bandwidth requirement for the router of 1.9
Mbps. The router must have some ability to select communication channels to avoid interference with the
RC transmitter, since they both operate at 2.4 GHz.
In addition, the router helps satisfy the following customer requirements: REQ 2.2.1.2.E.P1, REQ
2.2.1.3.A.P1, REQ 2.2.1.3.B.P1, REQ 2.2.1.3.C.P1, REQ 2.2.1.4.A.PI, REQ 2.2.1.4.B, and REQ
2.2.1.4.D.P1.
5.2.5.1.2
Alternatives
Router alternatives for this project include an AT&T 2Wire router73 and an Apple Airport
Express router.74 The two frequencies used by IEEE 802.11n are 2.4 and 5 GHz.
5.2.5.1.3
Decision Criteria
Since all commercial routers provide the necessary bandwidth and channel selection capabilities
outline above, the cost of the router, range, and time necessary for implementation were the major criteria.
The time necessary to get the routers implemented was found by first looking at the AT&T 2Wire router
and attempting to set it up as a network, which was not allowed by a custom firmware running on it.
After two hours attempting to change the AT&T router settings, the Apple router was tested and set up in
less than one hour in order to fulfill the requirements.
For a final implementation of this project, 5 GHz would be considered because it would not
interfere at all with the 2.4 GHz RC transmitter.
5.2.5.1.4
Decision
Since the AT&T 2Wire router had a custom firmware that would have needed to be modified for
the project, the Apple Airport Express router was used since it was possible to set up for the needs of the
project.
5.2.5.1.5
Implementation
The router is located in the vehicle for this project. This allows the base station to connect to the
VGPU onboard the quadcopter. Static IP addresses are set up for the VGPU and the laptop in order to
make video communication simpler to implement for the prototype version, where only one instance of
the product is in use. Static IP addresses were chosen so that the VGPU knows the base station
computer’s IP address is unchanging and can therefore send information to the same location every time.
85
5.2.5.1.6
Unit Test
The router was tested by connecting multiple devices to it and running the video stream as well as
streaming the telemetry data. The router passed this test, as it was able to simultaneously provide
connection to all of the devices while streaming both telemetry and video information during full
integration tests.
86
6
System Integration and Testing
This chapter discusses the system integration tests the team performed to ensure the system was
working correctly. It is divided into function sections of telemetry streaming, video streaming, GPS
tracking, RC control via RC interface module, and tracking algorithm. In addition, these functional
sections are again divided according to the team’s iterative building philosophy.
6.1
Telemetry Streaming
This was a test to see if full communication could be achieved from the quadcopter to the base
station. This test was performed in several stages. The first step was to pass a message from the
quadcopter, through the GUI, and into the RC interface module. Second, the team attempted to pass an
actual telemetry message (with GPS coordinates) from the quadcopter, through the base station computer,
and to the RC interface module. The last test attempted to quantify the latency experienced with
communication and evaluate it.
6.1.1 Test: Full System Communication
This was the first test of telemetry streaming to see if messages sent from the quadcopter could be
passed through the GUI and received on the RC interface module.
6.1.1.1
Equipment
 Base station computer
 RC interface module
 Quadcopter
6.1.1.2
Steps
After unit communication tests had been performed on the respective subsections, the system was
tested by attempting to pass a message from the quadcopter through the base station computer to the RC
interface module. In order to view the results, the RC interface module then sent the message back to the
base station computer to be viewed in a debugging text box. These tests can be seen in Appendix A.2.
6.1.1.3
Units of Measurement
This was a pass / fail test.
6.1.1.4
Definition of Success
If the information sent from the quadcopter is received on the base station computer, then the test
is a success. A full pass of this test is if the data sent from the quadcopter is passed through the base
station computer and received on the RC interface module correctly.
6.1.1.5
Results
This test was performed several times. After working through several bugs, such as setting the
correct baud rates and standardizing the message passed, the team achieved test success. The next step
was to perform a system communication test with live telemetry data.
6.1.2 Test: Communication with Live Telemetry Coordinates
This test took the previous test step further by actually sending live telemetry data. This tested
various parsing functions in both the base station computer and RC interface module to ensure that they
selected the correct data variables.
87
6.1.2.1
Equipment
 Full system
6.1.2.2
Steps
The quadcopter sent a specific data message with GPS information and telemetry data over
TCP/IP. The base station computer received this message and forwarded onto the RC interface module
via serial port. Again, the RC interface module sent the message back to the computer for viewing and
debugging purposes in the GUI.
6.1.2.3
Units of Measurement
This was a pass / fail test.
6.1.2.4
Definition of Success
This test would be a success if telemetry information seen was identical to the message sent from
the quadcopter.
6.1.2.5
Results
Initial testing proved that team members did not agree on the order of variables to be passed.
Some individuals passed, latitude as the first item in the message, while others passed longitude as the
first variable. This miscommunication was fixed and subsequent tests proved that a standardized message
was being correctly sent and received.
6.1.3 Communication Latency
The final communication test was one that measured communication latency. This testing used
the timestamps of the GPS modules to measure the latency of the system.
6.1.3.1
Equipment
 Full system
6.1.3.2
Steps
The system communication was initialized. Among other things, the time gathered from the base
station’s GPS (Adafruit) was compared to the time gathered from the quadcopter GPS (MediaTek). This
difference was reported back to the team via base station computer GUI debugging boxes as well as
debugging files. Communication latency was investigated in the tests from section A.2 in the Appendix
and mentioned in section A.5.3.2 of the Appendix.
6.1.3.3
Units of Measurement
The unit of measurement was time difference in seconds.
6.1.3.4
Definition of Success
With a top speed requirement of 4.2 m/s (REQ 2.2.1.2.B.P1), a tracking tolerance of 7.2 m (REQ
2.2.1.2.C.P1), and a worst case of 6 m offset due to GPS modules, the telemetry streaming latency could
𝑚
only account for a maximum of 1.2 m or error at the top speed. 1.2𝑚/4.2 𝑠 gives a maximum delay of
0.3 seconds. Streaming telemetry data with latency lower than 0.3 seconds is considered a success.
6.1.3.5
Results
Initial tests showed the latency could take ten seconds to stabilize and then was roughly 1-2
seconds delayed. This was unacceptable. The team implemented serial flushes on the GPS serial ports to
88
prevent any queuing of data. A subsequent test shows nearly immediate stabilization of latency; however,
it was stable around 0.9 seconds. In order to improve this latency, the team changed the refresh rates of
both GPS modules from their default speeds of 1 Hz to 10 Hz. This allowed the system to communicate
much faster since the data received was much faster. A final test with this new speed increase showed
average latencies of 0.2 seconds.
6.2
Video Streaming
The video streaming tests performed quantify the performance of various video streaming
protocols. Tests were performed to determine video latency, as well as video quality.
6.2.1 Video Latency
This test helps quantify the latency of each video streaming protocol.
6.2.1.1
Equipment
 Base station computer
 Quadcopter
6.2.1.2
Steps
With telemetry and video streaming running, the team used a stopwatch to quantify the video
latency in seconds. The camera pointed at the stopwatch and a picture was taken of the stopwatch and the
screen to which the video was being displayed. The difference between the time on the stopwatch and the
time depicted on the screen showed the latency of communication. Figure 43 shows this process. Section
A.6.1 in the Appendix shows the full latency test.
6.2.1.3
Unit of Measure
The unit of measure was time in seconds to determine the video latency. This test can be seen in
section A.6 of the Appendix.
6.2.1.4
Definition of Success
A successful video latency test results in a measured video latency of less than or equal to one
second as described in requirement 3.2.1.4.D.P1.
6.2.1.5
Results
RTSP experience a latency of about 1.8 seconds. This test was repeated with GStreamer yielding
a latency of about 0.2 seconds. Lastly, FFmpeg was tested and a video latency of 1.7 seconds was
observed. Figure 43 shows the results of this test. GStreamer was clearly the best video protocol
concerning latency.
89
RTSP
GStreamer
FFmpeg
Figure 43: Video Latency Results
6.2.2 Video Quality
With latency tested, the team wanted to compare each video stream to their respective video
quality. If GStreamer turned out to have far worse quality than the others did, its lower video latency
would be outweighed by its poor quality.
6.2.2.1
Equipment
 Base station computer
 Quadcopter
6.2.2.2
Steps
The team flew the quadcopter and streamed telemetry data in order to inspect the video quality
during an actual flight, as that is when the video is used. This test can be seen in section A.6.2 of the
Appendix.
90
6.2.2.3
Units of Measure
Four types of errors were used to classify video quality: Type 3, 2, 1 and 0.5. The type numbers
correlate to full screen blurs, partial screen blurs, minimal screen blurs, and video jitter respectively.
Each error type was weighted. Type 3, 2, 1, and 0.5 were weighted 15, 10, 5, and 2 points respectively.
A high score is bad as it shows that particular video protocol had more errors and thus worse quality.
Choosing the type number for a flaw was difficult at times, but having two team members watch the
stream and agree on the errors helped keep the tests consistent.
6.2.2.4
Results
Figure 44 shows the result of this test. GStreamer proved to be the best protocol in terms of
video quality. A recognized weakness of this test is that we knew which video stream had the lowest
latency and therefore could potentially bias the results in favor of that protocol. The next iteration of this
test would have multiple unbiased parties view the video stream and rank the flaws in order to eliminate
this potential bias.
Figure 44: Video Protocol Quality Scores
6.3
GPS Tracking
The tracking tests are used to slowly expand and develop the autonomous tracking algorithms.
The first two tests deal with GPS calculations in the microcontroller. The results from this calculation
determine where the quadcopter is located with respect to the base station and where it needs to fly next.
After this, the remaining tests inspect the RC commands being sent to the quadcopter. These are slowly
integrated with the GPS position results to develop the full autonomous tracking algorithms.
6.3.1 Stationary GPS Precision
The first GPS test compares the GPS positions of the GPS modules. Both modules are in the
same location, so in theory the calculated differences should be zero meters. Several different tests were
performed, many in different arbitrary locations, to help increase variety of the testing environments.
6.3.1.1
Equipment
 Base station
 MediaTek GPS
 VGPU
91
6.3.1.2
Steps
For the stationary GPS system precision tests, both GPS modules were placed in the same
location in a car. For the stationary tests, the car did not move while data was recorded. This was done to
confirm REQ 2.2.1.2.C.P1. See section A.5.3 of the Appendix for full results
6.3.1.3
Units of Measurement
Distance offset in meters was used to determine how accurate the GPS modules were.
6.3.1.4
Definition of Success
Since the GPS modules were placed in the same location, the distance offset should be zero
meters. The test was considered a success if the difference was reported within a 7.2 m radius as seen in
Figure 2 per REQ 2.2.1.2.C.P1.
Table 25 below shows the results of a different stationary test. While the longitude was well
within the bounds, the latitude most certainly was not. One possible source of errors was that this
location was close to buildings and featured lots of tree cover. It is possible that one or both GPS
modules were not able to obtain an accurate GPS fix. Cases like these were handled on an individual
basis so that the team could learn from them and improve system performance.
Table 25: Another GPS Stationary Result (10 Hz)
Time (s)
Latitude (m)
Longitude (m)
Average
0.2
9.07
1.40
Standard Deviation
0.05
3.28
1.13
6.3.2 Moving GPS Precision
The moving GPS precision test expanded upon the stationary test. Both GPS modules were again
in the same vehicle, however now the vehicle was moving. This test, while checking the calculated
difference between the GPS positions, also helped the team realize the importance of improving
communication latency. The initial moving test was done at 1 Hz refresh rates, and then later repeated
with the GPS modules operating at 10 Hz. An important realization was made during these tests. In
order to meet REQ 2.2.1.2.C.P1, the GPS modules have to be synchronized with very low latency. While
latency does not affect stationary tracking, it has a major effect when the modules are moving.
6.3.2.1
Equipment
 Base station
 MediaTek GPS
 VGPU
6.3.2.2
Steps
Moving GPS precision tests were performed many times. The first test was with a 1 Hz GPS
refresh rate, and later tests with a 10 Hz GPS refresh rate. These tests are located in Appendix A.5.3.1
and A.5.3.2, respectively. For these tests, the MediaTek GPS module, VGPU, and base station were
placed less than one foot apart inside a vehicle. Once a connection had been established, the team began
driving. GPS differences were saved to a file on the base station computer for post-test processing. This
process was done for many different routes.
92
6.3.2.3
Units of Measurement
Distance offset in meters was used to determine how accurate the GPS modules were.
6.3.2.4
Definition of Success
This test was an informative test about the accuracy of our two GPS modules. Because we were
only looking for information from this test, there was no definition of success, but instead, lessons learned
about GPS accuracy in different situations as well as the effect of telemetry data latency on system
performance.
6.3.2.5
Results
In one test, the team drove at speeds around 22.35 m/s (50 mph) and gathered the average
longitude and latitude differences. For this particular test, the team was driving north and south;
therefore, they expected to see greater latitude differences than longitude differences. Because the time
𝑚
difference was roughly 0.7 seconds, (0.7𝑠 ∗ 22.35 = 15.6 𝑚) this means that GPS data could be off by
𝑠
15.6 m just because of communication latency. This is an unacceptable range, and it pushed the team to
increase the refresh rates of the GPS to reduce this large threshold. Table 26 below shows the data
results.
Table 26: Moving GPS Precision Test (1 Hz)
Time (s)
Latitude (m)
Longitude (m)
Average (104 points)
0.73
8.81
3.50
Standard Deviation
0.13
10.59
1.54
Figure 45 and Table 27 below shows the results for the test repeated at the 10 Hz refresh rate.
Figure 45: Moving GPS Precision Test (22.35 m/s, 10 Hz)
93
Table 27: Moving GPS Precision Test (22.35 m/s, 10 Hz)
Time (s)
Latitude (m)
Longitude (m)
Average (1036 points)
0.20
6.27
3.95
Standard Deviation
0.04
4.51
2.20
The team was pleased with the results given the car could travel four meters at these speeds with
a 0.2 second latency. In addition to the tolerance due to communication latency, the six-meter tolerance
still exists due to GPS specifications. Combined, this gives a ten-meter window due to tolerances. This is
over half of the radius of the previous window with 1 Hz refresh rate. About 56% of the data recorded
fell inside the radius requirement of 7.2 m.
In order to test in the most realistic situation the team tested GPS precision while the modules
were moving at speeds on average at 16.9 m/s (38 mph) through various city streets. Multiple stoplights
were encountered during the test causing speeds to vary from a complete stop to 20.1 m/s (45 mph).
Figure 46 and Table 28 below show the results of this test.
Count
28th W --> E Paris N --> Burton W
-9.96
Latitude Difference
Longitude Difference
4.36
1.25
-3.11
-6.23
Figure 46: Moving GPS Precision Test (16.9 m/s, 10 Hz)
Table 28: Moving GPS Precision Test (16.9 m/s, 10 Hz)
Time (s)
Latitude (m)
Longitude (m)
Average (1779 points)
0.2
3.86
1.84
Standard Deviation
0.05
3.01
1.73
The results of this data showed that 88% of the data points fell within the 7.2 m radius. This was
a good result. The interesting fact is that direction of travel changed multiple times, as well as the speed.
This was definitely the most realistic of the GPS precision tests performed.
94
Nevertheless, the team was pleased with the results gathered. The 10 Hz refresh rate on the GPS
modules greatly reduced communication latency and thus increased GPS precision. With the GPS data
proven, the team was able to eliminate a potential source of error in the tracking algorithm.
6.4
RC Control via RC Interface Module
The team performed flight tests to iteratively determine how the system behaved when controlled
via base station through the RC interface module. These tests were not used with any GPS or telemetry
feedback from the communication loop.
6.4.1 RC Pass Through
This test ensured full manual control was maintained through the RC transmitter even when it
was communicating with the quadcopter via the RC interface module.
6.4.1.1
Equipment
 Full system
 Safety net
6.4.1.2
Steps
A program was created to give a full pass through ability to the RC transmitter. This means the
quadcopter should fly just as if the RC transmitter were not connected to the RC interface module. To
start, the RC values being received by the flight controller were viewed using the open sourced MW GUI.
Once these were verified, the quadcopter was flown and controlled using the basic maneuvers: pitching,
rolling, yawing, and increasing throttle. Team members held the safety net underneath the quadcopter in
case throttle needed to be cut to stop erratic flight behavior.
6.4.1.3
Units of Measurement
This was a pass / fail test determined by inspection.
6.4.1.4
Definition of Success
A successful test was reached once the quadcopter was flown to a hover and controlled as if a
regular transmitter was used.
6.4.1.5
Results
Initial tests with the MW GUI were conducted using a purchased logic converter to control the
voltage steps both up and down. This test resulted in a non-linear step up and step down of the values.
Once the design decision in section 5.2.2.1.2 was made, the test was successful, showing that full control
of the quadcopter from the base station microcontroller was possible.
6.4.2 Pitch and Roll Recording
This test was used as a means for gathering the pitch and roll potentiometer data coming out of
the RC transmitter. In order to determine what roll and pitch input voltages corresponded to actual
quadcopter flight, manual control was used, and the pitch and roll voltages were recorded during specific
flight maneuvers. This data was incorporated into the tracking algorithm to autonomously control the
quadcopter in a stable manner.
6.4.2.1
Equipment
 Full system
 Safety net
95
6.4.3 Steps
The team created a program to record RC values during RC pass through as mentioned in the
prior section. The quadcopter was flown to a hover while the pass through function was running;
enabling the microcontroller to echo the exact RC input values. Once a physical switch on the transmitter
was activated, the program would read that switch through a digital input pin and print out the values of
the movement in question (pitch or roll). These values were analyzed to get a rough estimate of the
appropriate values for the follow program.
6.4.3.1
Units of Measurement
This test was pass/fail, seeing as it would record data or it would not.
6.4.3.2
Definition of Success
A successful flight test was one where the quadcopter was controlled to pitch and roll at certain
speeds and the values being sent through the microcontroller were recorded for later analysis.
6.4.3.3
Results
The result of this test was that the quadcopter behaved as expected, and the pitch and roll values
were recorded for later use.
6.4.4 Autonomous Pitch and Roll Start
With pitch and roll values recorded, the team needed to test the values on the quadcopter to
ensure that the autonomous movement was controlled and stable. The test used to show this is described
below.
6.4.4.1
Equipment
 Full system
 Safety net
6.4.4.2
Steps
The quadcopter was flown to a hover at roughly two meters high, but this time when the physical
switch on the RC transmitter was flipped, the program would take control of the axis of movement that
was under inspection. This test was performed for both pitch and roll independently. Team members
held the safety net underneath the quadcopter in case throttle needed to be brought down on the
transmitter for an emergency landing.
6.4.4.3
Units of Measurement
This test was a pass / fail test, whether or not the quadcopter responded as expected.
6.4.4.4
Definition of Success
A successful test was one that the quadcopter slowly began accelerating on the axis of test until
the team members could no longer keep up with the quadcopter.
6.4.4.5
Results
When the quadcopter responded as expected, it was considered a success. It took several
iterations of this testing in order to tweak the output values for roll and pitch to achieve the desired
response. With this test complete, the team could incorporate this part of the autonomous control into the
tracking algorithms. The video of the autonomous pitch test can be found here.
96
6.4.5 Autonomous Pitch Stop
Since the tracking algorithm chosen would cause the quadcopter to be moving at its fastest speed
when entering the 3-meter radius around the base station, a stop function was needed for cases where the
base station was stationary. This function reverses pitch for a prescribed amount of time in order to stop
within a small distance.
Equipment
 Full system
 Safety net
6.4.5.1
Steps
The team wrote a program such that when the switch was activated, the stop function would
control the direction of movement and bring it to a stop. This was verified in the MultiWii GUI to verify
correct system response before actual flight-testing. This initial test verified that the test would not have
any adverse effects on the quadcopter, which could cause it to become unsafe. Next, the quadcopter was
manually hovered at about two meters high and the pilot started the quadcopter moving forward. Once
the switch was activated, it called the stop function. Team members held the safety net underneath the
quadcopter in case throttle needed to be cut for an emergency landing.
6.4.5.2
Units of Measurement
This test was measured in meters.
6.4.5.3
Definition of Success
A successful test would be one that showed the quadcopter stopping in six meters or less.
6.4.5.4
Results
Since the purpose of the stop function is to bring the quadcopter stop, the test was run multiple
times until a full stop was achieved without starting backward movement of the quadcopter. The final
stopping function yielded a stopping distance of 3 m +/- 1 m, well within the required distance of six
meters. The video of the pitch stop test can be seen here.
6.5
Tracking Algorithm
With telemetry streaming, roll and pitch controls verified, and stop functions finished, the
tracking algorithm was realized. By combining all of these subsystems, tests were performed to ensure
that they worked together correctly, and resulted in the correct system response. These tests ranged from
reading RC values to actually flying the quadcopter and running the program.
6.5.1 RC Response Test
Before testing the tracking algorithm with a flying quadcopter, a basic test was executed to verify
the RC interface module algorithm was sending the correct RC commands based on GPS locations.
6.5.1.1
Equipment
 Full system
 Equipment cart
 Vehicle
97
6.5.1.2
Steps
The full loop of communication was initiated without streaming video, and the base station was
powered from a parked vehicle. Next, the quadcopter was placed on a movable cart with a computer
connected to the flight controller. The computer was used to inspect the RC values via the MultiWii GUI.
To test the actions of the tracking algorithm, the cart with the quadcopter was moved to multiple positions
relative to the base station. These positions were determined in order to test the actions of every possible
combination of relative positions (i.e.: north, northeast, east, southeast, etc.). Once in each position, the
values of the RC receiver were analyzed for correctness. Full documentation for this test can be seen in
section A.4.1 of the Appendix.
6.5.1.3
Units of Measurement
This test was pass / fail.
6.5.1.4
Definition of Success and Results
Table 29 shows the positions tested, and the expected actions, which were all achieved. All
relative positions are greater than 7.2 meters away in order for the program to take appropriate actions.
Table 29: Expected Reactions to Position Changes
Relative Position of
Quadcopter
Desired Pitch
Actual Pitch
Desired Roll
Actual Roll
North
Lowered
Lowered
Centered
Centered
North East
Lowered
Lowered
Lowered
Lowered
East
Centered
Centered
Lowered
Lowered
South East
Raised
Raised
Lowered
Lowered
South
Raised
Raised
Centered
Centered
South West
Raised
Raised
Raised
Raised
West
Centered
Centered
Raised
Raised
North West
Lowered
Lowered
Raised
Raised
Within 7.2 m Radius
Centered
Centered
Centered
Centered
6.5.2 Motor Response Test
After verifying that the RC interface module responded by sending the correct RC values, the test
was performed on a fully implemented quadcopter. This was not a continuous flying test in the formal
sense, as quadcopter “hopping” was performed.
6.5.2.1
Equipment
 Full system
98
6.5.2.2
Steps
The quadcopter was moved to each possible position relative to the base station (i.e.: north,
northeast, east, southeast, etc.) and the throttle was increased to allow the quadcopter to fly just high
enough to verify it was pitching and/or rolling in the correct direction before bringing it back to the
ground. This type of flight “hopping” helped lower the risk of severe crashes in the event of incorrect
system response.
6.5.2.3
Units of Measurement
This test was pass / fail.
6.5.2.4
Definition of Success and Results
When each portion of the test was run, the quadcopter should tilt toward the base station when the
flight “hop” was performed. Although motor values were thought to be confirmed per earlier tests, the
quadcopter seemed to fly in the wrong direction several times. After testing by quadcopter hopping, the
quadcopter correctly navigated towards the desired location. From this point, the team could implement
their first complete tracking algorithm as discussed below.
6.5.3 North-Facing Flight Tracking
The first full tracking algorithm assumed the quadcopter always faced north, hence the “northfacing” naming convention. With all of the previous steps completed prior, a full flight test was
performed to see the results of this tracking algorithm. The stop function was removed from this test
since, once initiated, it never returned control of the quadcopter to the tracking algorithm. Because of
this, the quadcopter would drift with wind or momentum once either the latitude or the longitude data
came within the threshold. For this test, the GPS calculations were set at an offset of ten meters north of
the actual base station. This was done for safety reasons as multiple team members were standing at the
base station to initialize the test.
6.5.3.1
Equipment
 Full system
 Equipment cart
6.5.3.2
Steps
The communication loop sending telemetry from the quadcopter to the base station was initiated
while the quadcopter was on the ground. The quadcopter was then flown to a hover roughly ten meters
away from the base station and the program initiated via switch. Yaw was controlled manually in order to
keep the quadcopter facing north. Throttle was also manually controlled so that an emergency landing
could be performed at any time during the test. In the software that parses the telemetry and base station
position data on the RC interface module, the ground position was modified to be 10 m north of the actual
base station. This ensured that the quadcopter would fly at a given position that was not occupied by
team members and equipment.
6.5.3.3
Units of Measurement
Since the test was to see whether the quadcopter would track towards the base station, and correct
if it drifted out of bounds, this test was pass / fail.
6.5.3.4
Definition of Success/Results
A successful test was one in which the quadcopter continually flew toward the base station,
always correcting back towards the base station when it drifted out of the threshold. The quadcopter
99
tracked the base station within about 20 m, which was expected since a stop function was not being used.
More information on this test is found in section A.4.2 of the Appendix. A video of this test is available
here.
6.5.4 Orientation Tracking
The next iteration of tracking algorithm included yaw control. In this algorithm, the quadcopter
would yaw towards the base station and then begin pitching. As the base station moved, the quadcopter
would continue to yaw and pitch, so roll control was not needed in this new tracking method. Two major
tests were performed on this new form of tracking and both are described below.
6.5.4.1
Orientation Flight Test
The orientation flight test investigated the yaw control of the tracking algorithm. A
preprogrammed flight path was determined, and the quadcopter controlled via that path. These results
helped the team realize their second iteration of full tracking.
6.5.4.1.1
Equipment
 Full system
 Safety net
6.5.4.1.2
Steps
The software for the test was written for use without GPS positioning, but with the full
communication loop running in order to gather orientation data. The program was started as soon as the
quadcopter was hovering at around two meters, and a switch initiated by the user started the program.
The program initially ran as if it was ten meters south of the base station and began moving to the north.
Once its maximum acceleration was achieved, the program stopped the quadcopter and changed the
artificial position to ten meters east of the base station. Since the quadcopter at this point was facing
north, the orientation part of the program under test should turn the quadcopter to face west and then
accelerate forward. This process was repeated for three sides of a large square. The fourth leg of this
square was no conducted in order to ensure the quadcopter did not fly at the operators. The quadcopter
was then stopped and throttle was cut, causing it to fall into the safety net. The throttle in this test was
again in manual control mode. Team members held the safety net underneath the quadcopter in case
throttle needed to be cut for an emergency landing
6.5.4.1.3
Unit of Measurement
Since the test is based on whether or not the quadcopter turns to the appropriate direction before
accelerating, this test was pass / fail in nature.
6.5.4.1.4
Definition of Success/Results
A successful test is one where the quadcopter autonomously flew three sides of a box without
losing control or becoming unstable. This test was run successfully two times. The third test resulted in
the quadcopter hovering on the edge of the orientation threshold. This caused the quadcopter to drift
northeast instead of due north, and as a result the quadcopter hit a divider. The test was run a fourth time
with the same success as the initial runs. Figure 47 below shows the heading data during this final test.
As one can see, the quadcopter starts north, makes a banking turn towards the west, and finishes facing
south. More information on this test can be found in section A.3.3 of the Appendix, along with further
explanation on the crash. A video showing this box test is found here.
100
Figure 47: Orientation Flight Test
6.5.4.2
Orientation Flight Test with Tracking
This test combines the proven orientation above and adds it to the full telemetry communication
loop. This is a test of the second iteration of full tracking. With a stationary base station, the quadcopter
was tested to ensure it turned toward the base station and pitched forward. Then, the stop function was
initiated by the tracking algorithm to stop the quadcopter within the tracking threshold. Once the
stationary test was validated, the test was run again to verify that it worked with a moving base station.
6.5.4.2.1
Equipment
 Full system
6.5.4.2.2
Steps
To test the final orientation based tracking algorithm the team set up the communication on a
movable cart. Once this was established, the team walked around the base station with the quadcopter
mimicking flight while viewing RC input data through the GUI. Once this part of the test showed the
quadcopter was getting correct RC commands the team tried the test with a flying quadcopter. The
quadcopter was flown to a hover in a position approximately ten meters southeast of the base station, and
a physical switch on the RC transmitter activated the follow algorithm. Throttle was maintained as a
manual control in case the quadcopter needed to be brought down quickly. Once stationary tracking was
validated, the test was run from a vehicle, which was moving. This tested the moving base station case of
the follow program.
6.5.4.2.3
Unit of Measurement
This test was measured in meters between the base station and the quadcopter.
101
6.5.4.2.4
Definition of Success
A successful test was one in which the quadcopter would turn toward the base station and begin
slowly pitching toward it. Once over the base station, the follow program should call its pitch stop
function, causing the quadcopter to stop within 7.2 meters of the base station. Given then quadcopter
does not have a way of stopping drift due to wind or small amounts of momentum, a successful test could
drift out of the radius, causing the quadcopter to turn and pitch back to the radius.
6.5.4.2.5
Results
When the test was run, the follow program was first validated by walking with the quadcopter
and inspecting the RC values. This inspection showed that the quadcopter was trying to yaw to face the
base station. Once the quadcopter was turned to face the base station, it received commands to begin
pitching toward the base station. These successful results were enough to try the test with the quadcopter
flying. With the quadcopter hovering outside the radius, the switch was flipped, and the quadcopter flew
into the radius and an attempt to stop was observed. Due to wind, the quadcopter often drifted out of the
radius within 10 seconds or missed the radius completely, however the program responded appropriately
by turning the quadcopter around again and pitching toward the base station. The team considered this a
success. A video of the stationary test is shown here.
With a successful stationary case completed, the team set up the test in a vehicle and moved while
the tracking algorithm was running. The result was that the quadcopter continued to update the position it
was trying to reach as the base station moved. Like the prior iteration, the quadcopter often blew out of
the radius due to wind, but the quadcopter continued to adjust its pitch and orientation to overcome this
disturbance. A test video with the moving base station is available here.
102
7
Project Management
This chapter describes the organizational structure of the project. It shows how the team
organized project documents, design areas, management assignments, conflict resolution, and individual
safety protocol. In addition, this chapter compares predicted the work breakdown schedule (WBS) and
the budget with actual results. Finally, this chapter discusses the method of approach used by the team.
7.1
Documentation Organization
The team organized all of its documents into Google Drive, which is online real-time
collaboration software that is entirely cloud-based. It allowed the team to work on various documents
simultaneously from any location. The documents are organized into several main categories; Code,
Design Journals, Manuals, Media, Miscellaneous Notes, Planning Documents, Status Reports, Team 08:
Turn-In, and Tests. Only one document resides outside of these categories, and that is the Time Tracking
document. This is accessed frequently by the team and is where each team member records
time. Because of the frequency of access, it was not placed into a sub-category. Below, each of the main
categories is described.
7.1.1 Code
The Code folder is where the team kept all of the working software code for the project. This
provided an area from which each team member did revision control, keeping a record of software as it
changed throughout the course of the project. This also allowed all members to see the various source
code for consistency and brainstorming purposes.
7.1.2 Design Journals
Design Journals were the place where each team member kept a design area as they worked
through various portions of the project. Each team member was required to keep track of various
sketches and calculations as design work was completed. There was a separate folder for each team
member to keep this information. It also provided an audit trail to prove various design decisions along
the way.
7.1.3 Manuals
The Manuals section contains manuals for different components relating to the project. This
could be a manual for the flight controller on the quadcopter, documentation on the Raspberry Pi, or
information on the RC transmitter.
7.1.4 Media
The Media folder contains many various media that pertains to Team Eagleye. This includes a
wide variety of videos during testing throughout the year. It also holds the team posters, concept art, and
each of the presentations that were given during the year.
7.1.5 Miscellaneous Notes
Miscellaneous Notes is a place where team members placed anything that does not seem to fit
anywhere else. This section holds different brainstorming that the team did on finding a solution. This is
also the area where notes were kept on meetings with the team’s faculty advisor and industrial consultant.
7.1.6 Planning Documents
The Planning Documents section holds the majority of documents the team has created. Version
control was implemented such that any modifications done to a reviewed document are done to a new
revision of the document name. This more clearly defines different revisions than those provided by
103
Google Docs’ automatic revision history. This also helped the team keep a baseline history of past
revisions in the event that they need to be revisited. This folder also holds the team budget, block
diagrams, decision matrix, and feature matrix. It is also where various sub-sections of the Project
Proposal Feasibility Study (PPFS) and Design Report were created and initially edited. Lastly, it holds
the group’s WBS and Gantt chart.
7.1.7 Status Reports
Weekly status reports were created in this area and emailed to our faculty advisor. These status
reports include hours for each person during the week, for the entire project, as well as the group’s
running total of hours. The reports also include tasks completed, goals for the next week, along with any
issues that have hindered progress and may result in the need for outside help.
7.1.8 Team 08: Turn-in
The Turn-in folder is the main folder shared with our faculty advisor. It is where documents were
placed by the team for official review. Any document placed in here only returned to its original
destination as a new revision. This ensured that edits are not made on documents currently awaiting
review; it also was a critical part in implementing version control.
7.1.9 Tests
“Tests” is the file location where the team placed various tests that were performed throughout
the year. This included both unit tests as well as integration tests. Each test has a specific folder that
contains the raw data, charts, and graphs in a spreadsheet, and a summarizing file. This file explains the
test that was performed, displays the results, and includes links to the source code that was used during
the test.
7.2
Team Organization
This section describes each team member’s design areas and management tasks. In addition, this
section also establishes guidelines for how the team is to collaborate, design, and maintain safety. The
team organization was done in such a way that each member had specific technical and managerial
assignments.
7.2.1 Technical Assignments (Design Areas)
The project was been split into six different design areas. Each team member was assigned as the
primarily lead on at least one design areas, and as a secondary lead for at least one different area. The
amount of design areas assigned to an individual was directly related to the relative size of design work in
each area. The size of design work was estimated by what the team thought the complexity of each piece
was. This changed throughout the year as areas became larger or smaller in scope then originally
predicted. Having secondary members on a design area allowed for better cohesion within the group. It
also helped with task management and troubleshooting, as the secondary person provided support and
helped when a design area was behind or experiencing difficulties. These design areas are seen in Table
30 and the individual design areas are detailed below.
104
Table 30: Design Areas and Team Member Assignments
Design Area
Primary
Secondary
Base Station
Components
Quadcopter Components
Quadcopter
Drew
Robert
RC Transmitter
All
Telemetry
Data from
Quadcopter
Drew
Spencer
Base Station Computer,
Wireless Router
Flight Controller, GPS
Module, VGPU
Tracking
Algorithm
Robert
Spencer
Base Station Computer,
RC Interface Module,
GPS Module
None
RC
commands
Robert
Spencer
RC Interface Module,
RC Transmitter
RC Receiver
GUI
Jared
Spencer
Base Station Computer,
RC Interface Module,
Wireless Router
VGPU
Video
Streaming
Spencer
Jared
Base Station Computer,
Wireless Router
VGPU, Camera
7.2.1.1
Quadcopter
The quadcopter design area incorporated all flight aspects of the quadcopter. This included
getting the quadcopter assembled and calibrating the flight controller in order to achieve stable flight.
Other aspects of this design area included crash repairs, physical modifications needed for any reason,
and the mounting of additional hardware. This section required all components of the quadcopter, as well
as the RC transmitter from the base station.
7.2.1.2
Telemetry Data from Quadcopter
This section included sending GPS coordinates and other telemetry data from the flight controller
and GPS module through the VGPU. From here, it is sent to the base station computer using the network
provided by the wireless router. The base station computer must receive this data and convert it into a
usable form.
7.2.1.3
Tracking Algorithm
Once the base station computer receives the GPS position and telemetry data from the
quadcopter, it is sent to the RC interface module. This is where the tracking algorithms reside. The
tracking algorithm then sends its corresponding data to the RC commands design area.
7.2.1.4
RC Commands
The tracking algorithm output tells the RC interface module what action is needed, and the RC
commands design area handled the conversion of flight commands to RC signals. Once this conversion
was complete, they were sent over RC to the RC receiver on the quadcopter.
105
7.2.1.5
GUI
The GUI design area included displaying both the video stream and telemetry data. It handled all
of the communication between the VGPU on the quadcopter and RC interface module. In addition, it
provided an interface from which a user can interact with the system, as well as a debugging interface for
testing the system.
7.2.1.6
Video Streaming
This design area worked with both the VGPU and camera as well as the base station computer.
The best video streaming protocol was found such that it met all previously defined requirements. This
design area required communication between the base station computer and VGPU over the network
provided by the wireless router.
7.2.2 Management Assignments
Jared Zoodsma: Financial Manager - Maintained records of financial expenditures and budget
status. All part orders were made through Jared in order to avoid duplicate or incompatible
components being ordered.
Robert Hoff: Marketing Manager - In charge of all media and public aspects of group work.
Duties included in marketing were posters, website updates, social media, and graphical diagrams
requiring computer software.
Drew Brandsen: Professionalism Manager - In charge of maintaining document formatting and
professional look. Drew proofread team documents before turning them in so that consistency
was maintained between all documents.
Spencer Olson: Team Manager - In charge of keeping track of deadlines set by the faculty
adviser, as well as intermediate deadlines. He was also responsible for turning in completed
assignments and aiding overall team direction.
7.2.3 Working Guidelines
As seen above, the project has been broken into several design areas. Each member of the team
was assigned at least one primary design area and other secondary design areas, as extra assistance was
needed. If a component or task fell into one of these design areas, it was the primary investigator’s
responsibility to ensure the task was completed. The primary investigator asked for assistance from the
secondary investigator, or others if need be, but the direct responsibility fell on the primary investigator.
Frequent meetings were held to ensure the team was up to date on members' current tasks. In
addition, a spreadsheet was used to keep track of team hours. Every time a team member entered time, he
also classified the type of work (design or administrative), included what area he worked on, what was
accomplished, and what will be worked on next. This was another way the team stayed accountable and
current on what other members were working on.
In the event a team member did not meet a deadline, the team member was expected to complete
work in order to make up for the short falling. If the deadline was a stretch goal that has little to no effect
on other components, the repercussions were minimal. If deadlines were continually missed, a meeting
would have been held with all team members with the importance of making deadlines being stressed.
Thankfully, throughout the year all team members were hard working and met most deadlines on time.
No one person was consistently delaying the team causing the need for a team meeting. In addition, the
team reviewed the workload for all team members to ensure all members have reasonable expectations
and made appropriate changes when necessary.
106
7.2.4 Safety Guidelines
The team had a number of guidelines that provided limits on how work is done. When flying
without the tether, two team members needed to be present. This was for both safety and debugging
reasons. If a team member gets hurt while flying the quadcopter it is important to have someone else
there to help. In the event of a crash, it helped to have more than one set of eyes to see the crash. Since
each person may have differing opinions on what the crash can affect, having an extra set of eyes made
the debugging of errors much smoother after crashes.
7.2.5 Team Communication and Accountability
The team met weekly in order to discuss overall direction and scheduling. These meetings helped
to foster communication between team members, ensuring that all members meet expectations and know
what needed to be completed next. An hourly record of member project work was also recorded to ensure
that it was obvious if a team member is pulling too much, or not enough, weight in the project. Team
members hold each other accountable for putting in considerable time every week in order to keep the
project moving, but also understand when personal circumstances arise.
7.3
Schedule and Work Breakdown Schedule
The following section details the schedule of the project and discusses major milestones
completed throughout the project.
7.3.1 Hours Summary
During the course of the year, the team kept track of what time was spent on what areas of the
project. This was done using a spreadsheet online. The team also developed a WBS to estimate and
guide the schedule of design and administrative tasks. This estimate includes a contingency factor of 1.5
to help mitigate unplanned delays and missed tasks. A list of the hour breakdowns is seen in Table 31
below.
Table 31: Summary of Hours
Design Time
Admin Time
Total Time
773 hours
416 hours
1189 hours
Project Estimate (1.5x)
1159.5 hours
624 hours
1783.5 hours
Project Actual
1115.75 hours
839.0 hours
1954.75 hours
43.75
215.0
171.25
Project Estimate
Difference
As you can see, the team has completed a total of 1115.75 design hours out of 1159.5 hours
estimated, or about 43.75 hours below estimate. Administrative hours prediction was low as the team
went 215.0 hours over the estimate of 624 hours. Overall, the team is 171.25 hours above the estimated
total hours. However, this is only 9.6% higher than the predicted hours, which is very good for a
yearlong project. A cumulative graph of hours worked this year is seen in Figure 48. The red area shows
the design hours and the blue area shows administrative hours. The team is happy with the mixture of
administrative hours and design hours. A complete copy of the team’s WBS is found on the team’s
website.
107
Figure 48: Graph of Hours Worked on Project
7.3.2 Milestones
Major milestones observed in this project are seen below.
 Project Selection and Approval
 October 14 - First Presentation
 October 17 - Streaming Video
 November 13 - Stable flight of Prototype 0.1
 November 19 – Operational Budget
 December 2 - Second Presentation
 December 6 - Full control of Prototype 0.1
 December 9 - PPFS
 January 22 - Prototype 0.2 flying, basic RC control through Arduino achieved
 February 21 - Basic GUI, telemetry data successfully gathered from MW
 March 5 - Third Presentation
 March 14 - Base Station under construction, flying with video stream
 April 5 - Prototype 1.0 assembled and flying
 April 11 - Autonomous tracking of stationary Base Station
 April 15 - Orientation added to autonomous tracking
 April 25 – CEAC Review
 May 5 - Fourth Presentation
 May 5 – Faculty Review
108
 May 10 - Final Presentation and Senior Design Open House
 May 15 - Final Design Report
7.4
Operational Budget
In order to manage funds and stay within a reasonable financial range, the group used a budget to
project costs and track funds appropriately. The budget estimation is compared to the actual
funds that were spent. A summarizing section at the end discusses some of the discrepancies that
occurred as well as the final resulting expenses.
7.4.1 Project Costs
Table 32 is the team’s operational budget. On the left hand side are the budgetary projections as
of December 9, 2013. It includes a contingency factor of 1.5 to anticipate failures, crashes, and
unanticipated costs. On the right hand side of the table are the actual unit prices and quantities purchased.
The final column shows the net difference between the two.
Table 32: Team 08: Eagleye Budget
Budget Estimation as of 12/9/2013
Part
Unit Price
Actual as of 5/15/2014
Quantity Unit Price
Quantity
Difference
Flight Controller75
$160.00
1
$65.36
1
$94.64
35
$40.00
2
$39.99
1
$40.01
76
$20.00
5
$13.44
5
$32.80
Set of Four Propellers77
$10.00
3
$9.77
7
-$38.40
59
$25.00
1
$34.11
1
-$9.11
Raspberry Pi78
$35.00
1
$35.00
1
$0.00
79
$50.00
1
$45.01
1
$4.99
80
$15.00
1
$9.99
2
-$4.98
XBee Antenna81
$20.00
2
$0.00
0
$40.00
82
$15.00
5
$17.71
7
-$48.97
RC Transmitter83
$30.00
1
$34.97
4
-$109.88
84
$80.00
1
$19.99
1
$60.01
85
$50.00
1
$28.93
1
$21.07
Shipping Costs
$100.00
1
$113.55
1
-$13.55
Expandable Features
$100.00
1
$17.38
1
$82.62
Unplanned Quad Parts
$0.00
0
$92.86
1
-$92.86
Unplanned Base Parts
$0.00
0
$83.51
1
-$83.51
GPS Module
ESC
Raspberry Pi Camera
Arduino Due
Wi-Fi Dongle
Brushless Motors
Frame and Landing Gear
Cables and Connectors
C. Factor
Total:
1.5x
$1,455.00
109
Current Total:
$1005.26
Remaining:
$449.74
Percentage:
69.09%
7.4.2 Summary
The first thing one notices is that the team is currently under budget by nearly $450.00 (31%).
There are a number of reasons that this occurred. First is smarter component selection. During interim,
another flight controller was tested that proved equally capable as the $160.00 APM. This flight
controller saved the team nearly $100 in costs and included a GPS module, adding another $40.00 in
savings. The team also had a number of components donated by Professor Yoon Kim. These parts
included four motors and an aluminum frame and landing gear set. This saved the team another $150.
However, there were additional expenses in items such as propellers, RC transmitter, and other unplanned
quadcopter and base station components. The RC transmitter was a particular point of pain as the first
two transmitters were destroyed through their connections to the RC interface module. As such, the team
needed to order two more RC transmitters (one as backup). As a result, this particular component went
well over budget. However, through the donations gathered and wiser component selection, the team
showed great stewardship in the funds they were granted. However, for a real world project, going over
in hours causes the project to go over budget too. As section 7.3.1 highlights, the team went slightly over
the expected time for the project.
7.4.3 Sources of Funding
The team received two main sources for funding the project. One source of funding came from
the team’s corporate sponsor, DornerWorks. DornerWorks graciously offered to donate $1,000 towards
the team’s budget. Calvin Engineering Department’s Senior Design Fund provided the remaining $455
outlined in the budget above. This value was originally $726.50, however, the team was pushed towards
cutting down on the budget in order to meet the wider Senior Design budget for this academic year.
7.5
Method of Approach
The Method of Approach section describes how the team executed the design process. It also
discusses specific design norms that influenced design decisions, as well as research techniques used.
7.5.1 Design Methodology
When deciding on how a design should come together, the team used a strategy that combined
research and group brainstorming sessions. When a design decision was needed, the group recorded
various options and the effectiveness of each were subsequently researched and evaluated. Once the
group had ample time to research the topic, those who found the most pertinent material presented it to
the group in an informal manner and a decision was ultimately made.
As for the actual designing, the team employed agile development. This strategy encourages
small but positive design steps towards achieving the final design goal. This was especially evident in the
software design, which included many revisions as new features and bug fixes were added to the code.
7.5.2 Research Techniques
In order to find information that was relevant to topics under question, the team used a
combination of forums, informational websites, scientific journals, as well as professors on campus.
Depending on the topic, the team chose the most effective source, which usually depended on the size of
the design in question. For more broad design questions, research databases and scientific journals were
utilized, while smaller technical questions were easily answered by forums. One of the major goals of
this portion was to consider all relevant options including new options uncovered by research.
110
8
Business Plan
This chapter develops a comprehensive business plan around the project. Beginning with the
overview, the plan discusses the vision for the company, a current market study, competition, a business
model, and lastly the financial forecast. The next chapter will give a conclusion for the project by stating
major risks as well as analyzing the feasibility of the project.
8.1
Overview
Eagleye LLC is an engineering design firm that specializes in small, unmanned aerial vehicles
(UAVs). Eagleye was founded on the principles of trust, justice, and transparency. Eagleye LLC’s staple
product is a small autonomous quadcopter. This quadcopter is designed to capture aerial video and
stream the video to a remote location in real time. In addition, the quadcopter is capable of following a
ground vehicle autonomously. This product could be applied to various security situations such as aerial
reconnaissance over moving targets or highly maneuverable perimeter surveillance.
8.2
Vision and Mission Statement
The vision and principles of the company are listed below and describe the drive that is behind
the company.
8.3
Vision for the Company
Eagleye LLC is a company that strives to improve the protection of military and diplomatic
personnel by using UAV systems. The team understands the sacrifice troops and leaders make for this
country and its freedom. Eagleye LLC’s mission is to give back by providing them with one more layer
of security as they travel in hostile environments. This gives Eagleye’s work on UAV systems a sense of
justice and purpose.
8.4
Values and Principles on which the Business Stands
The products created by Eagleye represent its employees personally. Poor products reflect poorly
on those who designed and built them. As such, Eagleye strives for the highest quality in each of its
products and operations. Above all, the products produced must convey a sense of trust. If those using
Eagleye’s products cannot trust the information they are being given, the entire purpose of the quadcopter
product is lost. The other major principle is transparency. It must be inherently clear what information
the users are receiving from the system. In other words, the product must never be so difficult or
complicated to use that it actually decreases the awareness of the user. These principles guide Eagleye
LLC as its team designs and produces next-generation UAVs.
8.5
Marketing Study
Eagleye’s target market consists primarily of military and security companies. Although military
services most likely have some form of UAV or satellite services already, Eagleye attempts to fill a need
that is not currently met by existing UAVs. This product would fall between an individual's first person
ground view and an existing UAV or satellite flying thousands of feet in the air. It would also fall into a
price range that is not currently filled by the UAV market. The following sub-sections detail the market
size and growth, along with regulator restrictions, barriers to entry and exit, and lastly key success factors.
8.5.1 How Large is the Market?
The current market for military spending on UAVs is about 10.6 billion86 and for non-military
spending is near 6.6 billion87. This equates to a total market of around 17.2 billion dollars for both the
military and nonmilitary spending on UAVs.
111
8.5.2 Is it Growing or Shrinking?
The future of UAV technology looks bright, according to AIAA (American Institute of
Aeronautics and Astronautics), “non-military application of Unmanned Aerial Vehicle (UAV) technology
is the fastest growing manufacturing sector in the global aerospace industry and expected to grow by
700% between 2012 and 2018.”87 Furthermore, another article by Market Research Media continues to
say that it expects military UAV spending to be over $86.5 billion during the next five years.86 This
shows that both sides of the market (commercial and military) are rapidly growing and expected to
continue doing so.
8.5.3 Regulatory Restrictions
The UAV industry has been subject to significant regulation and restriction since military drone
strikes as well as privacy concerns have made it a fiercely debated topic. There is a perception among
Americans that UAVs represent a threat to citizen privacy and security; however, this is starting to
change. Increasingly more Americans are recognizing the vast potential uses for small UAVs in
nonmilitary markets such as farming.87 In fact, the United States Congress recently passed legislation, the
Federal Aviation Administration (FAA) Reauthorization Act, which will replace the ban on UAV use
with a set of commercial regulations by 2015.88 Because these regulations are not yet well established,
Eagleye LLC will carefully monitor the regulatory situation of the industry, but it is expected that the
predominantly untapped commercial small UAV market will quickly become both legal and profitable.86
There are two additional concerns related to regulatory restrictions, safety requirements, and military
standards. Due to the nature of the quadcopter product’s status as a small electronic vehicle with fastmoving parts, Eagleye LLC will be careful to comply with safety requirements as the FAA releases
them. It is expected that these requirements will be similar in nature to model plane, electronic device,
non-car vehicle, and privately owned airplane regulations. When considering products that are intended
for military use, Eagleye LLC will be sure to satisfy military requirements as they are stated in any
contracts secured; typically, military-grade products have very exact specifications and documentation
that is even more detailed on the design process.
8.5.4 Barriers to Entry and Exit
Several factors will create barriers to enter this market. First and most considerable is the large
upfront capital needed. Economies of scale will need to be utilized to amortize design costs over multiple
sales. In addition, Eagleye will need to work closely with the customers to achieve customer satisfaction
and eventually increase name recognition among potential customers.
8.5.5 Key Success Factors in the Industry
A key step towards success for the company will be exposing the overlooked niche this product
covers for security and military clients. Companies will not be willing to purchase new UAVs when they
do not see the need for them, so it is critical that companies are made aware of how this product can
bolster their security without requiring additional labor.
8.6
Competition
Competition is an important factor to consider when entering a market. The competition section
looks at both existing competitors in this market, as well as other companies that could enter this market.
8.6.1 Existing Competitors
Existing Competitors are listed below and all compete on some level with Eagleye LLC.
112
8.6.1.1
Hex AirBot
Hex AirBot manufactures a competing quadcopter product for $160; however, this quadcopter is
designed and marketed towards enthusiasts as a toy.89 Eagleye LLC will respond to this competitor by
including features for and focusing on a customer basis of commercial security companies. One crucial
feature Eagleye will include that Hex AirBot does not is GPS and vehicle-following abilities.
8.6.1.2
DJI
DJI is an active competitor in the small UAV market, offering four different copters between
$110 and $15000.90 Their target customer base is primarily enthusiasts and those who require airborne
video streaming.90 Eagleye LLC will respond to this competitor in two main ways. First, the product will
include features not offered by DJI’s cheaper models, such as GPS navigation and vehicle
tracking. Second, Eagleye LLC plans on designing and marketing its product for commercial security
companies as opposed to exclusively enthusiasts and media professionals.
8.6.1.3
FireBox
Firebox designs and supplies various electronic products; one of these is the MicroDrone
quadcopter for $120.91 The MicroDrone quadcopter is marketed towards enthusiasts as a toy for
primarily indoor use.91 Eagleye LLC will respond to this competitor by focusing on the features that
Firebox failed to include, such as video capturing, outdoor flight optimization, and GPS navigation.
8.6.1.4
Lockheed Martin
Lockheed Martin is an active player in the larger military drone market with peripheral activity in
the commercial UAV market. They manufacture many unique models of UAVs of all sizes.92 They
could represent a significant threat to Eagleye’s market niche with some of their models, such as the
quad-vtol UAV; however, they too are more heavily focused on military contracts with very expensive
drones than they are on commercial or private security UAVs.92 Lockheed Martin appears to dedicate
much more of its resources to the R&D of new products than it does to the expansion and mass
production of its current ones.92 Eagleye LLC will respond to this competitor by focusing on commercial
and private security applications for its significantly cheaper quadcopter product.
8.6.1.5
Walkera
Walkera designs and supplies collectible RC helicopter and quadcopter products. These products
are focused on aesthetics and impressive flight patterns, ranging from $70 to $500.91 Eagleye LLC’s
quadcopter product is significantly different from Walkera’s models because of its video streaming, GPS
navigation, and vehicle tracking features. Eagleye’s product will also be designed for functionality and
robust operation, as opposed to solely aesthetics, which will help them target commercial security
companies instead of RC enthusiasts.
8.6.2 Potential Competitors
Potential competitors listed below all have the capability to enter this market and present
a viable threat to Eagleye LLC.
8.6.2.1
Northrop Grumman
Northrop Grumman is an active player in the larger UAV market; they manufacture several
different drone models. The smallest of Northrop Grumman’s UAVs has a 10-foot wingspan, which is
still significantly larger than Eagleye LLC’s staple product.93 It is also important to note that the majority
of their models are planes, not quadcopters, and that they focus almost exclusively on military
customers.93 If Northrop Grumman were to build smaller, cheaper, more locally useful products, they
113
could enter the same market niche as Eagleye LLC. This is unlikely because it would represent a very
significant change in their target customers and Northrop Grumman currently possesses military
contracts, which tend to be reliable source of business.
8.6.2.2
AII Corp
AII is another current competitor in the larger UAV market; they manufacture a few different
drones of varying sizes. All of their drones are plane, not quadcopter styled, and they focus nearly
exclusively on expensive military and defense applications for their UAV products.94 If AII were to begin
producing UAVs that were smaller and cheaper, they could enter Eagleye LLC’s market niche, but this is
unlikely, due to the stable nature of their current military customers.
8.6.2.3
SightLogix
SightLogix is a provider of perimeter security technology such as cameras and sensors.95 This
company is interesting because they could be either a potential customer or a potential competitor. If they
decide to offer aerial quadcopter surveillance products, they could outfit a quadcopter from Eagleye LLC
with their cameras and sensors, or instead develop a competing quadcopter product. Eagleye LLC will
ensure that its products are high quality enough to both encourage the purchase of their product by
commercial security companies and discourage the in-house development of competing products by those
companies.
8.7
Business Model
The business model shows how the company plans to operate. It includes the desired market
image, company goals, a SWOT (strengths, weaknesses, opportunities, threats) analysis, as well as a
competitive strategy.
8.7.1 Desired Image and Position in the Market
Eagleye desires to be a highly respected producer of small-scale UAVs for security and military
purposes. This market niche is nearly empty and ready for new innovative players.
8.7.2 Company Goals and Objectives
The company goals below briefly describe various objectives in both operations, finances, and in
other areas.
8.7.2.1
Operational
Eagleye LLC hopes to become a highly innovative company that uses latest technology to design
and build UAVs for military and other security customers.
8.7.2.2
Financial
Eagleye must be a profitable venture. Using the first prototype as a showcase, the company
hopes to gain further funds to create a robust final prototype that can be used to sell the idea of a
quadcopter with these features and abilities to interested customers.
8.7.2.3
Other
Eagleye hopes to inspire a company culture of transparency and collaboration to help spur
creativity and continuous improvement amongst employees.
8.7.3 SWOT Analysis
The SWOT Analysis for Eagleye LLC is described below. It discusses the internal strengths and
weaknesses of the company, along with external opportunities and threats.
114
8.7.3.1
Internal Strengths
A major strength of Eagleye is its leaders; they are people who strive for quality and understand
that the work they complete reflects on them. Another strength is the product itself. It has highly
expandable set of unique features that will set Eagleye apart from other producers of UAVs.
8.7.3.2
Internal Weaknesses
A weakness that is apparent with Eagleye LLC is a lack of market experience and low name
recognition. Since Eagleye is a startup, its sales representatives will need to have strong confidence in
their products and use extensive knowledge of both the market and customer in order to win additional
business. In addition, it is important to ensure Eagleye produces highly reliable products to gain customer
confidence.
8.7.3.3
External Opportunities
Primarily, Eagleye LLC hopes to move into a market niche that is not highly developed. Smallscale UAVs can exhibit high quality while being much more cost effective compared to larger wellknown UAVs, like the Predator Drone, which runs close to 4 million dollars per vehicle.96 Eagleye LLC
hopes to exploit these differences in building and marketing their products for less money than the
competition.
8.7.3.4
External Threats
One threat, which Eagleye LLC recognizes, is the presence of competing companies in similar
niches in the UAV manufacturing and design market. Unlike Eagleye LLC, the majority of these
companies, such as Northrop Grumman and Lockheed Martin, are well established with very expensive
products. These companies have been producing successful products for decades, developing trust
between them and their customers. Since Eagleye is a startup, the company will need to provide a
flawless product up front in order to begin building similar trust with customers.
8.7.4 Competitive Strategy
A brief competitive strategy is described below to show how Eagleye plans to address
competition.
8.7.4.1
Differentiation
Eagleye plans to compete using product differentiation. Because the systems designed are small
scale UAVs for security systems, they are much smaller than what is on the market today. This allows
Eagleye to operate at a significantly lower price point than most UAV manufacturers. Technology
improvements in the last decade have made this possible by integrating many features into a much smaller
package. By opening up this new market niche, Eagleye hopes to provide a new solution to customer
security needs.
8.7.4.2
Focus
The initial focus will be demonstrating the usefulness and possibilities of small-scale UAVs, as
well as encouraging the association of outstanding quality and trust with company products. It will be
imperative to build trust with customers right away in order to continue competing in this market. Once
brand name recognition and customer trust have been established, focus will shift to providing a greater
portfolio of UAVs to allow for even more applications, further increasing the market share and
diversifying the customer base.
115
8.8
Financial Forecasts
The financial forecasts for Eagleye LLC are detailed below. The forecasts include Pro-Forma
Income Statements and Cash Flow statements for the first three years. Also included are the Break-Even
Analysis and a Ratio Analysis. Each of these statements is derived from supporting calculations that are
seen in the next section.
8.8.1 Key Assumptions
This financial analysis rests on several key assumptions including operation without stockholder
investment, 10% annual interest rate on bank loans, even distribution of cash flows, no delayed purchases,
and capacity-demand assumptions. Operating Eagleye LLC without stockholder investment is an
assumption that should be very accurate, as the leadership does not wish to make Eagleye public. The
effects of operating without stockholder investment are a more limited initial capital with less
accountability and leverage. Assuming a 10% annual interest rate on bank loans is an assumption that is
as accurate as research allows. Realistically, this number will deviate from 10% but there is no
guaranteed prediction of future interest rates. The annual interest on bank loans affects the financial
situation of Eagleye LLC by defining how much money must be paid for each dollar
borrowed. Additionally, for this financial analysis it is assumed that all sales and costs are distributed
evenly throughout the year in which they are recorded. This assumption makes calculations far simpler
and should not significantly change final values. Finally, Eagleye LLC assumed that the company does
not have any accounts receivable or payable within their first three years.
The key assumptions that Eagleye LLC is making in regards to capacity and demand involve their
ability to produce quadcopters and their ability to sell those quadcopters. For the purposes of this
financial analysis, it is assumed that Eagleye LLC will successfully manufacture 10, 200, and 400
quadcopters in each of their first three years respectively. This assumption has a far lower manufacturing
quantity for the first year because that time will be spent primarily in product design and manufacturing
plant startup. Furthermore, this assumption shows increasing manufacturing quantity as time progresses
because of anticipated increase in customer relations and demand. A final assumption that Eagleye LLC
is making is that all of its produced quadcopters will be sold within the year in which they were
produced. This is intended to simplify calculations, and the assumption’s accuracy is dependent upon the
industry’s current competitive and customer situation. A complete view of supporting calculations for the
following financial statements is seen in Table 33 below. This table breaks down fixed and variable costs
for each of the first three years.
116
Table 33: Supporting Calculations
8.8.2 Financial Statements
Two financial statements were used to analyze the financial footing of Eagleye LLC. These are
the Pro-Forma Income Statement and Cash Flow Statement. They are described in the following sections
with the actual statement included.
8.8.2.1
Income Statement
The income statement shows the various revenues and costs for each of the first three years. This
statement can be seen in Table 34 below. The most important number in this statement is the net income
(or net loss) seen at the bottom of the table. As is seen, the first year results in a net loss of over half a
million dollars. This is anticipated, as there are a lot of startup costs and further design work to get the
product to production. By year two, the net income is over one million and year three results in a net
income of over 2.6 million dollars.
117
Table 34: Income Statement
8.8.2.2
Cash Flow Statement
The cash flow statement, which is shown in the Table 35, highlights the major areas where cash
will be spent and received. In this section the loan information, as well as the plan to pay it off are listed,
showing the three-year pay off period. In addition, it shows the current value of the loan at each
year. This section also highlights the purchase of and depreciation on equipment.
Table 35: Cash Flow Statement
118
8.8.3 Break-Even Analysis
Break-even analysis is detailed in Table 36 and shows a break-even point for sales of 55, 55, and
54 units in the first three years respectively. This break-even point was calculated for a unit cost of
$15,000. Although the anticipated production in year one is only ten units, Eagleye anticipates surpassing
the break-even point in years two and three, making up for lost ground.
Table 36: Break-Even Analysis
8.8.4 Ratio Analysis
The Ratios used can be seen at the bottom of this section in Table 37. Note that all ratio values
are negative for year one. This is because Eagleye LLC is operating at a loss while it incurs setup costs
and before it begins collecting revenue. The debts to assets ratio in years one, two, and three are -39.18%,
26.91%, and 0% respectively. The goal was to pay off debt as quickly as possible to decrease liabilities.
The total asset turnover is negative for year one, because Eagleye is not selling a significant number of
quadcopters. Eagleye starts selling in year two and this ratio goes up to 4.61 times. After year two
Eagleye pays off all debt and begin accumulating the extra cash (assets) that would go to paying interest
on debt. For this reason, the total asset turnover goes down to 1.92 times in year three. Again, the profit
margin is very poor in year one because the company does not have any sales; however, it gains nicely to
37.31% and 44.54% in years two and three respectively. The same is true for the gross margin of revenue
that goes from -7.37% in year one to 81.27% and 83.65% in year two and three. The reason this ratio is
so high once Eagleye starts selling units is because of a fairly large markup on the product.
119
Table 37: Ratio Analysis
8.9
Summary
In conclusion, Eagleye LLC plans to provide high quality and highly reliable UAVs for security
purposes. We feel there is a unique market niche for our product since existing UAV systems can cost
millions of dollars. Our low cost and unique product will distinguish us in the growing market of
autonomous surveillance. UAV surveillance is a multi-billion dollar market that is only expected to
grow. As can be seen in the cost estimation sections above, we believe this is a viable business venture
and worth pursuing.
120
9
Conclusion
This chapter summarizes the project in terms of the final project results. This is broken down into
the overall results, costs, and time spent. In addition, the team gives recommendations on the next steps
for future work that could be completed. Lastly, it ends with a summary for the project as a whole.
9.1
Project Results
The final results of the project are broken down into three categories; overall, final costs, and total
time spent. Each of these categories is explained below.
9.1.1 Overall
The team was able to build a quadcopter system that flew stably. A base station was also
assembled that included the necessary displays, computation, and communication equipment to fly the
quadcopter autonomously. The team achieved complete communication throughout the system and
provided useful information to the user in the GUI. Video streaming and map information was also
implemented in the GUI in order to provide the situational awareness required for the user. Lastly,
autonomous tracking with orientation was achieved. This was a significant milestone, and the team is
happy that they reached all of their major goals for this project.
9.1.2 Final Costs
The team acquired $455 from Calvin’s engineering department and a $1000 donation from
DornerWorks. The team was very concerned about the budget since the beginning, knowing that any
moment during flight tests a crash could set the team back several hundred dollars. With this in mind, the
team took several actions in an effort to reduce the risk of running over budget. First, the team used a
flight controller that was significantly less expensive than the one budgeted for, yet still performed as
needed. In addition, the team acquired several donations courtesy of Professor Yoon Kim and Justin
TeBrake. Finally, the team set contingencies in the budget for unplanned items and expandable features.
This careful planning helped the team stay under budget
9.1.3 Time Spent
The team spent a total of 1425 hours on the project. This was 120% greater than their estimated
time. Looking back on it, if a contingency factor of 2.0 had been used instead of 1.5, the estimation
would have been much closer. The hours completed, however, show a dedication to the project by all
members and their commitment to see it to completion.
9.2
Future Work
Despite the successes of this project, there is still work to be done. Given more time, the team
would like to rebuild the quadcopter with a commercial frame, higher-end flight controller, and higherend components. Additional sensors could also be added to improve the flight performance such as sonar
and an optical flow sensor. These steps would help increase precision of flight (especially in wind),
stability of the quadcopter, and the overall tracking.
The base station could also see some improvements as well. A custom board could be made that
would replace the Arduino-filter combination. A new 5 GHz Wi-Fi router would cut down interference
with the 2.4 GHz RC transmitter. Another option would be to use a lower frequency RC transmitter like
915 MHz. These hardware changes would help make the communication more consistent and reliable.
The software is likely the area that could see the most improvement. There are a lot of checks
and fail-safes that could be added to make the software much more robust and deterministic in the corner
cases of use. Before production, the team would want to know exact latencies of functions. One would
121
need to create system timers to find the latencies of the software. For example, one could time how long
it takes to receive TCP/IP information, filter it, and send it to the RC interface module in the GUI. The
autopilot code can always be tweaked to improve tracking as well. Additional methods of tracking like
IR could be added in order to improve tracking accuracy and response time, especially when the
quadcopter has line of sight to the vehicle. There are also improvements that could be made on the video
and telemetry streaming areas to improve quality and amount of information gathered. Lastly, new
features could be added to the GUI in order to increase system interaction and provide better data to the
user.
A necessary feature for full autonomy is obstacle avoidance. This was outside of the scope of our
project, but is necessary for a production model system. The team brainstormed a couple ways to
implement this, but never had time to research it more. One way the team brainstormed was to use sonar
sensors in front of the quadcopter. This would be a simple and quick addition to prevent the quadcopter
from running into objects. However, it would be difficult to know which direction to move in order to
avoid these objects. Another alternative would be to use a camera or several cameras and implement a
form of computer vision to detect objects and determine the best course to circumnavigate the object.
Adding obstacle avoidance would drastically increase the usefulness and market-readiness of this product.
9.3
Summary
Though this was an ambitious project, the Eagleye team showed the feasibility of this solution
through first semester research and preliminary designs. Next, the team designed a working prototype in
the second semester that met the project goals and did so while coming under budget. Autonomous
tracking of a vehicle was achieved by using a quadcopter as a platform. Video streaming from the
quadcopter showed the proof of concept for a product such as this in increasing the situational awareness
of the user(s). The team knew from the beginning of the year that the scope of their project was large;
however, they were dedicated to the project and worked closely with one another to ensure they met the
project goals. While there are always things to improve and redesign, the team is proud of their work
accomplished and is thankful to all those that helped them reach this goal.
122
REFERENCES
1
(n.d.). In Calvin College Engineering. Retrieved December 8, 2013, from
http://www.calvin.edu/academic/engineering
2
Lane (2013, November 7). In Wikipedia. Retrieved December 9, 2013, from
http://en.wikipedia.org/wiki/Lane#Lane_width_and_capacity
3
Wikipedia. (2013, November 13). Wikimedia Foundation. In City Block. Retrieved December 6, 2013, from
http://en.wikipedia.org/wiki/City_block.
4
Enhanced F Scale for Tornado Damage. (2007, February 1). In Storm Prediction Center NOAA. Retrieved
December 6, 2013, from http://www.spc.noaa.gov/faq/tornado/ef-scale.html.
5
USA.com. (2013). In Grand Rapids, MI Weather. Retrieved December 6, 2013, from http://www.usa.com/grandrapids-mi-weather.htm#HistoricalWind%20Speed.
6
Taoke, G. T. (1989, March). Brake Reaction Times of Unalerted Drivers. ITE, 19-21. Retrieved May 12, 2014, from
http://www.ite.org/membersonly/itejournal/pdf/jca89a19.pdf
7
Ergonomics: Lifting limits for employees. (2006, September 2). In Michigan.gov. Retrieved December 6, 2013,
from www.michigan.gov/documents/cis/wsh_lifting_176907_7.doc
8
Whittle, R. (2013, 09). Drone skies. In Popular Mechanics, 190, 76. Retrieved from
http://search.proquest.com/docview/1440248096?accountid=9844
9
FPV Quadcopter Shot Down By Radar. (n.d.). In Finkbuilt. Retrieved December 8, 2013, from
http://www.finkbuilt.com/blog/fpv-quadcopter-shot-down-by-radar/
10
Hobby Lobby (2013, December 8). In RC Helicopters | Beginner Heli | Collective Pitch from Hobby Lobby.
Retrieved from http://www.hobby-lobby.com/rc_helicopters_197_ctg.htm
11
RC Helicopter Selector (2013, December 8). In Skyartec Cessna 182 2.4 Ghz 5 Channel RC Airplane RTF
Brushless Version. Retrieved from http://www.rchelicopterselect.com/skyartec-cessna-182-2.4-ghz-5-channel-rcairplane-rtf-brushless-version.html
12
Squidoo (n.d.). In Radio Controlled Blimps. Retrieved December 8, 2013, from
http://www.squidoo.com/RemoteControlledBlimp
13
Blimpworks Airships (n.d.). In Blimpworks Airships. Retrieved December 8, 2013, from
http://rcblimp.net/blimp_outdoor.htm
14
Throwable Panoramic Ball Camera Captures 360 Shots When Sent Skyward (2013). In Steves Digicams.
Retrieved from http://www.stevesdigicams.com/news/throwable_panoramic_ball_camera_captures_360_shots_when_sent_skyward.html#b
15
ArduCopter Introduction (2012, November 17). In Google Project Hosting. Retrieved December 8, 2013, from
https://code.google.com/p/arducopter/wiki/ArduCopter_QuadIntroduction
16
NEWBIE, looking for long flight time. (2010, November 27). In RC Groups. Retrieved December 8, 2013, from
http://www.rcgroups.com/forums/showthread.php?t=1345539
17
Remote Control Airplanes Review 2014. (n.d.). In TopTenREVIEWS. Retrieved December 8, 2013, from
http://remote-control-airplanes-review.toptenreviews.com/
18
3DR RTF Quad (n.d.). In 3DRobotics Inc. Retrieved December 8, 2013, from
https://store.3drobotics.com/products/apm-3dr-quad-rtf
19
Align RC 6 Channel Helicopter 600 (n.d.). In Xheli. Retrieved from http://www.xheli.com/15h-kx0160npa.html
20
3DR ARF APM:Plane (2013). In 3DRobotics Inc. Retrieved December 8, 2013, from
http://store.3drobotics.com/products/3DR-ARF-APM:Plane
21
RC Blimp: Introducing the MicroBlimp (n.d.). In Microflight. Retrieved December 8, 2013, from
http://www.microflight.com/MicroBlimp-RTF-Set
123
22
Vendittelli, M. (2012). In Quadrotor Modeling. Retrieved December 8, 2013, from
http://www.dis.uniroma1.it/~venditt/didattica/eir/01_Modeling.pdf
23
Brain, Marshall, and William Harris. (2000, April 01). How Helicopters Work. Retrieved December 8, 2013 from
http://science.howstuffworks.com/transport/flight/modern/helicopter.htm
24
NMEA 0183. (n.d.). In Wikipedia. Retrieved December 8, 2013, from http://en.wikipedia.org/wiki/NMEA_0183
25
Frequency-hopping spread spectrum. (n.d.). In Wikipedia. Retrieved December 8, 2013, from
http://en.wikipedia.org/wiki/Frequency-hopping_spread_spectrum
26
Speed limits in the United States. (2013, August 12). Wikipedia. Retrieved December 8, 2013, from
http://en.wikipedia.org/wiki/Speed_limits_in_the_United_States
27
(n.d.). In NeverWet. Retrieved December 8, 2013, from http://www.neverwet.com/anti-wetting.php
28
Optical Flow Sensor. (n.d.).In ArduCopter. Retrieved December 8, 2013, from
http://copter.ardupilot.com/wiki/optical-flow-sensor/
29
Quandcopter control function layers. (n.d.). In HefnyCopter. Retrieved December 8, 2013, from
http://hefnycopter.net/index.php?option=com_content&view=article&id=22:quadcopter-control-functionlayers&catid=9&Itemid=103
30
Engelbert, Wenzel, K. E., Masselli, A., & Zell, A. (2011). Automatic Take Off, Tracking and Landing of a
Miniature UAV on a Moving Carrier Vehicle. Web: Journal of Intelligent & Robotic Systems.
31
Build a Miniature High-Rate Speed Control with Battery Eliminator Circuit (BEC). (n.d.). In Stefanv. Retrieved
December 8, 2013, from http://www.stefanv.com/rcstuff/escbec.htm
32
Explanation of the Brushless Motor / ESC System and Capacitor Function. (n.d.). In Traxxas. Retrieved
December 8, 2013, from http://traxxas.com/forums/showthread.php?8958604-Explanation-of-the-Brushless-MotorESC-System-and-Capacitor-Function
33
USB. (2013, July 12). Wikipedia. Retrieved December 7, 2013, from http://en.wikipedia.org/wiki/USB
34
Creating the camera board. (n.d.). In Raspberry Pi. Retrieved December 8, 2013, from
http://www.raspberrypi.org/archives/3525
35
Adafruit Ultimate GPS Breakout (n.d.). In Adafruit. Retrieved December 8, 2013, from
http://www.adafruit.com/products/746#Technical_Detail
36
AeroQuad 32 Flight Control Board Version 2. (n.d.). In AeroQuadStore. Retrieved December 8, 2013, from
http://www.aeroquadstore.com/AeroQuad_32_Flight_Control_Board_Version_2_p/aq32-001.htm
37
APM 2.5+ Assembled (Cables enter from top). (n.d.). In 3DRobotics Inc. Retrieved December 8, 2013, from
http://store.3drobotics.com/products/apm-2-dot-5-plus-assembled-set-top-entry
38
AutoQuad 6 Autopilot Flight Controller (n.d.). In Viacopter - Multicopters & Multirotors. Retrieved December 8,
2013, from https://viacopter.eu/multirotor-shop/flight-controllers/autoquad-6-multicopter-store
39
MWC MultiWii Lite Lightweight Version 4-axis Flight Control Board QUADX. (n.d.). Retrieved December 8,
2013, from http://www.goodluckbuy.com/mwc-multiwii-lite-lightweight-version-4-axis-flight-control-boardquadx.html
40
MultiWii 328P Flight Controller w/FTDI & DSM2 Port. (n.d.). In HobbyKing. Retrieved April 22, 2014, from
http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=35682
41
MultiWii PRO Flight Controller w/MTK GPS Module. (n.d.). In HobbyKing. Retrieved April 22, 2014, from
http://www.hobbyking.com/hobbyking/store/__26588__MultiWii_PRO_Flight_Controller_w_MTK_GPS_Module.
html
42
PX4FMU Autopilot / Flight Management Unit. (n.d.). In PX4 Autopilot Platform. Retrieved December 8, 2013,
from https://pixhawk.ethz.ch/px4/modules/px4f
43
UAVX-ARM32 Full Sensors. (n.d.). In QUADROUFO. Retrieved December 8, 2013, from
http://www.quadroufo.com/product_info.php?products_id=88&osCsid=d7lf5mikok4tplcs0lr1814qo3
44
Adafruit Ultimate GPS Breakout - 66 channel w/10 Hz updates. (2013). In adafruit. Retrieved April 21, 2014,
from http://www.adafruit.com/products/746
124
45
MultiWii PRO Flight Controller w/MTK GPS Module. (2013). In Hobbyking. Retrieved April 21, 2014, from
http://www.hobbyking.com/hobbyking/store/__26588__MultiWii_PRO_Flight_Controller_w_MTK_GPS_Module.
html
46
Estimate Electric Motor and Prop Combo. (n.d.). In adamone . Retrieved December 8, 2013, from
http://adamone.rchomepage.com/calc_motor
47
NTM Prop Drive Series 28-26A 1200kv / 286w. (n.d.). In HobbyKing. Retrieved April 22, 2014, from
http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=39435
48
TURNIGY Plush 25amp Speed Controller. (n.d.). In HobbyKing. Retrieved April 22, 2014, from
http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=2163
49
LiPoBatteries (2013). In Wikispaces. Retrieved December 8, 2013, from
http://quadcopter.wikispaces.com/LiPo+Batteries
50
Robotics beta (2012, November). In StackExchange. Retrieved December 8, 2013, from
http://robotics.stackexchange.com/questions/554/quadcopter-lipo-battery-weight-capacity-trade-off
51
Turnigy 2450mAh 3S 30C Lipo Pack. (n.d.). In HobbyKing. Retrieved May 12, 2014, from
http://hobbyking.com/hobbyking/store/__10274__Turnigy_2450mAh_3S_30C_Lipo_Pack.html
52
Turnigy 3000mAh 3S 30C Lipo Pack. (n.d.). In HobbyKing. Retrieved May 12, 2014, from
http://hobbyking.com/hobbyking/store/__9497__Turnigy_3000mAh_3S_30C_Lipo_Pack.html
53
Turnigy 3600mAh 3S 30C Lipo Pack. (n.d.). In HobbyKing. Retrieved May 12, 2014, from
http://hobbyking.com/hobbyking/store/__9505__Turnigy_3600mAh_3S_30C_Lipo_Pack.html
54
Turnigy 5000mAh 3S 30C Lipo Pack. (n.d.). In HobbyKing. Retrieved May 12, 2014, from
http://hobbyking.com/hobbyking/store/__9515__Turnigy_5000mAh_3S_30C_Lipo_Pack.html
55
2.4G CT6B 6-Channel Transmitter+Receiver (R6B). (n.d.). In HobbyPartz. Retrieved April 22, 2014, from
http://www.hobbypartz.com/79p-ct6b-r6b-radiosystem.html
56
(n.d.). In Raspberry Pi. Retrieved December 8, 2013, from http://www.raspberrypi.org
57
Products (n.d.). In Arduino. Retrieved December 8, 2013, from http://arduino.cc/en/Main/Products
58
BeagleBone (n.d.). In BeagleBone.org. Retrieved December 9, 2013, from
http://beagleboard.org/Products/BeagleBone
61
(n.d.). In PCB Quote Online. Retrieved April 22, 2014, from http://www.bittele.com/pcb-quote-online_T.asp
62
H.264/MPEG AVC. (n.d.). In Wikipedia. Retrieved December 8, 2013, from
http://en.wikipedia.org/wiki/H.264/MPEG-4_AVC
63
Real Time Streaming Protocol. (2014, May 4). In Wikipedia. Retrieved May 12, 2014, from
http://en.wikipedia.org/wiki/Real_Time_Streaming_Protocol
64
CSharp. (2010, March 22). In VideoLAN Wiki. Retrieved May 12, 2014, from https://wiki.videolan.org/C_Sharp/
65
(2014, April 20). In GStreamer. Retrieved April 23, 2014, from http://gstreamer.freedesktop.org/
66
(2014, May 1). In FFmpeg. Retrieved May 12, 2014, from http://www.ffmpeg.org/
67
Psips (n.d.). In GitHub. Retrieved April 23, 2014, from https://github.com/AndyA/psips
68
Raspberry Pi Camera Board. (n.d.). In Adafruit. Retrieved May 12, 2014, from
http://www.adafruit.com/products/1367?gclid=CJDshtzEp74CFckWMgodO1QADQ
69
HERO3 White Edition. (2014). In GoPro. Retrieved May 13, 2014, from http://gopro.com/cameras/hd-hero3-whiteedition?gclid=CLvpkaenqr4CFahaMgodp24AaQ
70
Genius 3220032100 WideCam F100 USB 2.0 WebCam (n.d.). In Newegg. Retrieved December 8, 2013, from
http://www.newegg.com/Product/Product.aspx?Item=N82E16826179091
71
GlobalSat BU-353 USB GPS Navigation Receiver (n.d.). In Amazon. Retrieved December 8, 2013, from
http://www.amazon.com/GlobalSat-BU-353-USB-Navigation-Receiver/dp/B000PKX2KA
72
BU-353 GPS Receiver (n.d.). In USGlobalSat. Retrieved December 8, 2013, from
http://www.usglobalsat.com/store/download/62/bu353_ds_ug.pdf
73
2Wire Gateway User Guide. (2008, June). In Pace. Retrieved May 12, 2014, from
http://knowledgebase.pace.com/view/766/2700_User_Guide.pdf
74
AirPort Express 802.11n (1st Generation). (n.d.). In Apple. Retrieved May 12, 2014, from http://support.apple.com/kb/sp1
75
APM 2.6 Set (external compass) (n.d.). In 3DRobotics. Retrieved December 8, 2013, from
http://store.3drobotics.com/products/apm-2-6-kit-1
125
76
TURNIGY Plush 25amp Speed Controller. (n.d.). In HobbyKing. Retrieved April 25, 2014,
fromhttp://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=2163
77
APC Propellers 10X47 Push-Pull Set. (n.d.). In 3DRobotics. Retrieved April 25, 2014,
fromhttps://store.3drobotics.com/products/apc-propeller-set-10x47-sfp-style
78
Raspberry Pi Model B 512MB RAM (n.d.). In Adafruit. Retrieved December 8, 2013, from
http://www.adafruit.com/products/998
79
Arduino Due - assembled - Due (n.d.). In Adafruit. Retrieved December 8, 2013, from
http://www.adafruit.com/products/1076
80
Raspberry Pi WIFI Adapter / Dongle - The Pi Hut (n.d.). In Amazon. Retrieved December 8, 2013, from
http://www.amazon.com/Raspberry-Pi-WIFI-AdapterDongle/dp/B009FA2UYK/ref=sr_1_6?ie=UTF8&qid=1386475850&sr=8-6&keywords=wifi+dongle
81
XBee 1mW Wire Antenna - Series 1 (802.15.4). (2010, December 13). In Amazon. Retrieved April 25, 2014,
fromhttp://www.amazon.com/XBee-1mW-Wire-Antenna-802-154/dp/B004G4ZHK4/ref=pd_sim_pc_1?ie=UTF8&refRID=187364R9TRBMVJBBVSD4
82
NTM Prop Drive Series 28-26A 1200kv / 286w (short shaft version) . (n.d.). In HobbyKing. Retrieved April 25,
2014, fromhttp://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=39435
83
Turnigy 6X FHSS 2.4ghz Transmitter and Receiver (n.d.). In Hobby King. Retrieved December 8, 2013, from
http://www.hobbyking.com/hobbyking/store/__24969__turnigy_6x_fhss_2_4ghz_transmitter_and_reciever_mode_2
_.html
84
Turnigy Talon Quadcopter (V2.0) Carbon Fiber Frame. (2014). In HobbyKing. Retrieved April 25, 2014,
fromhttp://www.hobbyking.com/hobbyking/store/__22781__Turnigy_Talon_Quadcopter_V2_0_Carbon_Fiber_Fra
me_550mm.html
85
Cables (n.d.). In 3D Robotics: UAV technology. Retrieved December 8, 2013, from
https://store.3drobotics.com/t/parts/cables
86
Haldane, M. (2013, August 8). U.S. Slowly Opening up Commercial Drone Industry. In Reuters. Retrieved
December 8, 2013, from http://www.reuters.com/article/2013/08/08/us-usa-drones-commercialidUSBRE97715U20130808
87
Projected Growth for Non-Military Use of Drones Presents New Business Opportunities (n.d.). In Business
Opportunities Journal. Retrieved December 8, 2013, from http://www.boj.com/projected-growth-for-non-militaryuse-of-drones-presents-new-business-opportunities/
88
Smithson, S. (2012, February 7). Drones over U.S. Get OK by Congress. In Washington Times. Retrieved
December 8, 2013, from http://www.washingtontimes.com/news/2012/feb/7/coming-to-a-sky-near-you/?page=all
89
Hex Airbot. (n.d.). In Hex Airbot. Retrieved December 8, 2013, from http://hexairbot.com/
90
DJI | All Products. (n.d.). In DJI. Retrieved December 8, 2013, from http://www.dji.com/products/
91
Micro Drone (n.d.). In Firebox. Retrieved December 8, 2013, from http://www.firebox.com/product/5565/MicroDrone?aff=512
92
Five Lessons for a Successful STEM Career. (n.d.) In Lockheed Martin. Retrieved December 8, 2013, from
http://www.lockheedmartin.com/us/products/procerus/quad-vtol.html
93
Unmanned Systems. (n.d.). In Northrop Grumman. Retrieved December 8, 2013, from
http://www.northropgrumman.com/Capabilities/Unmannedsystems/Pages/default.aspx
94
Unmanned Systems. (n.d.). In AAI Corporation. Retrieved December 8, 2013, from
http://www.aaicorp.com/products/unmanned-systems
95
Thermal Cameras and Video Analytics for Perimeter Security. (n.d.). In SightLogix. Retrieved December 8, 2013,
from http://www.sightlogix.com/
96
Drew, C. (2009, March 16). Drones Are Weapons of Choice in Fighting Qaeda. In New York Times. Retrieved
December 8, 2013, from http://www.nytimes.com/2009/03/17/business/17uav.html?pagewanted=all&_r=0
126
APPENDIX
Appendix A : Test Documents...................................................................................................... 2
A.1
Battery ............................................................................................................................... 2
A.1.1
A.2
Battery Life Test 1 ...................................................................................................... 2
Communication ................................................................................................................. 3
A.2.1
Communication Latency Test 1 .................................................................................. 3
A.2.2
Communication Latency Test 2 .................................................................................. 4
A.3
Field Test .......................................................................................................................... 5
A.3.1
Test 1 ........................................................................................................................... 5
A.3.2
Test 2 ........................................................................................................................... 6
A.3.3
Box Test ...................................................................................................................... 7
A.4
RC ................................................................................................................................... 12
A.4.1
Reactions for Following ............................................................................................ 12
A.4.2
Reactions for Following with Flight ......................................................................... 13
A.5
Sensors ............................................................................................................................ 14
A.5.1
Barometer .................................................................................................................. 14
A.5.2
Compass .................................................................................................................... 20
A.5.3
GPS ........................................................................................................................... 46
A.6
Video ............................................................................................................................... 81
A.6.1
RTSP Latency ........................................................................................................... 81
A.6.2
Summary ................................................................................................................... 85
A-1
Appendix A: Test Documents
A.1
A.1.1
Battery
Battery Life Test 1
Name: Drew Brandsen, Jared Zoodsma
Date: 3/15/2014
Test Number: 1
Item under test: Battery Life
Testing Parameters:
 2450, 2650, 3000, 3300, 5000 mAh batteries were tested
 Raspberry Pi was powered during test
 No telemetry or video streaming was done during the test
Notes:


Each battery was tested individually
A fully charged battery was placed in quadcopter it was flown until the quadcopter could no
longer be lifted due to low battery
Results:


There was a fairly linear relation between battery size and flight time
From this data, we can calculate that the system is drawing between 22 and 24 amps.
A-2
A.2
A.2.1
Communication
Communication Latency Test 1
Name: Jared Zoodsma
Date: 3/11/2014
Test Number: 1
Item under test: Communication Latency
Testing Parameters: (Ex. software version “xyz” used)
 GUI software used
 Arduino Code
 Pi Telemetry Code
 MultiWii 328P used
Notes:


Both GPS modules were placed on west-facing overhead door in engineering building
Test was done at 2:40 PM
Results:
 Took 10.4 seconds to get “caught up”
o Once the system was “caught up” there still seemed to be a disparity between GUI
system time and time reported from the GPS modules (more on this later)
 Although the clock on the computer was synced right before the test, there seems to be a
discrepancy between times in the GUI and from the GPS modules. This can be observed because
the difference between Quad time and GUI time is greater than the difference between Quad time
and Arduino time (1.71 seconds vs. .80 seconds). This is unexpected because the information
arrives at the GUI before it is passed to the Arduino.
o At first glance this is attributed to unsynced clocks on the devices. Another explanation
could be that the time is not updated in the code promptly.
A-3
A.2.2
Communication Latency Test 2
Name: Robert Hoff, Jared Zoodsma, Drew Brandsen
Date: 4/2/2014
Test Number: 1
Item under test: Communication Latency
Testing Parameters:
 GUI code
 Pi code
 Arduino code
Notes:



Base Station (Arduino, GUI) were placed by west overhead door
Pi with Multiwii-328P were placed by overhead door
Wi-Fi router placed by overhead door
Results:
 Erroneous latitude differences were reported ~7.5% of the time
o This data was reported to be ~50,000km latitude difference
o This issue seemed to go away when GPS modules were moved farther away from the EB.
This could mean GPS modules did not have very good fixes
o A work around could be to have a threshold. If differences are greater than 1km then
discard that data.
 Latencies bounced between 0.8 and 1.0 seconds
o Average of 0.9 seconds
A-4
A.3
A.3.1
Field Test
Test 1
Name: Drew Brandsen, Robert Hoff, Spencer Olson, Jared Zoodsma
Date: 2/18/2014
Test Number: 1
Item under test: Communication from quadcopter to base station, and preliminary tracking algorithm
and video streaming.
Testing Parameters:
 GUI version “2/19/2014” used on the Gateway
 RTSP video streaming protocol used (code in GUI)
 Arduino Code
 Telemetry Rev D with separate GPS serial port on Raspberry Pi
 Raspberry Pi used was one from the Calvin Lab (not team’s Pi).
 Pi Camera
 Adafruit GPS was used on base station
 APM GPS was used on quadcopter
 MultiWii 328p Flight Controller
 Car power adapter used to power the Eagleye network.
Notes:





Sunny weather with clear skies (GPS had good fix)
Used Jared and Spencer’s cars.
Quadcopter in one car and base station in another.
3 LEDs were used to display close, medium, and far distances between base station and
quadcopter in the direction of North, South, East, and West.
One car remained stationary while another car moved to various locations
o This was done with both the quadcopter remaining stationary and then again with the
base station remaining stationary
Results:
 GUI struggled rendering full HD picture.
 Range of video stream was about 66 feet (rough estimate).
 When the quadcopter car was stationary, the tracking algorithm worked well, but when the
quadcopter car was moving, there was latency in the data being received on the Arduino.
o This may have been due to a GPS data delay in the raspberry pi (data was queued and so
old data was being sent). This was amended by flushing the GPS serial port within the
data loop. Upon restarting the communication this issue was mostly resolved. Data was
then being sent within a second of real-time.
A-5
A.3.2
Test 2
Name: Spencer Olson, Jared Zoodsma, Drew Brandsen, Robert Hoff
Date: 3/03/2014
Test Number: 2
Item under test: Communication from quadcopter to base station, and preliminary tracking algorithm.
Testing Parameters:
 GUI version used on the Gateway
 RTSP video streaming protocol used (code in GUI)
 Arduino Code
 Telemetry Rev E on Raspberry Pi
 Raspberry Pi used was one from the Calvin Lab (not team’s Pi).
 Adafruit GPS was used on base station
 MediaTek GPS was used on quadcopter (came with MultiWii)
 MultiWii Pro Flight Controller
 Bestek car power adapter used to power the Eagleye network and GUI computer
Notes:



Cloudy Weather; GPS still had good fix when in the open
Used Drew and Rob’s Cars
Quadcopter with Raspberry Pi and camera in Drew’s Car. Base station with Wi-Fi Router in
Rob’s Car
Results:
 Found the Raspberry Pi was not receiving enough current to enable all feature (SSH) (when
plugged into USB adapter in car)
 Determined a Serial flush between GUI and Arduino was needed once TCP connection had been
established to decrease latencies
 Discovered reported distances (60m) were larger than actual distances (20m) when both cars were
traveling (major latency issues)
A-6
A.3.3
A.3.3.1
Box Test
Summary
Name: Drew Brandsen, Jared Zoodsma, Spencer Olson, Robert Hoff
Date: 4/15/2014
Item under test: Arduino Orientation Tracking Algorithm
Testing Parameters:
 Arduino Software
 GUI Software
 Telemetry Software
Notes:




Quadcopter was to fly a predetermined box pattern
o Fly north for a few seconds, turn west and fly west for a few seconds, turn south and fly
south for a few seconds.
Initial Test used a +/-30 degree threshold window
o if within +/-30 degrees of desired heading, begin pitching forward
Later Tests used a +/-10 degree threshold window
Expected heading data would be 0 → -90 → -180
Conclusion:
 Three out of four times the quadcopter correctly flew the pattern designated.
o One time we ran out of room in the TNT and the quadcopter hit the divider between the
tennis courts and the track.
 A smaller heading window should be explored to maximize accuracy of pitch movements.
 The tests were considered a success and will be integrated into GPS tracking to further improve
our tracking algorithms.
A-7
A.3.3.2
Box Test Number 1: +/-30 Degree Heading Window
Notes:


First test of orientation tracking
Quadcopter performed excellently
o Pattern was more of a semi-circle than a box due to the large heading threshold window
Results:
Summary:
 Good data seen here - quad holds fairly straight, then consistently yaws towards destination
heading. Once at desired heading, again holds steady
 There is a very gradual transition from North to South (during which the quadcopter is pitching
slowly forward).
o This is likely due to the large pitch threshold for the heading data (+/- 30 degrees)
A-8
A.3.3.3
Box Test Number 2: +/-10 Degree Heading Window
Notes:

Quadcopter performed great again
o Pattern was much more a box due to the smaller heading threshold window
Results:
Summary:
 Good data seen here - quad holds fairly straight, then consistently yaws towards destination
heading. Once at desired heading, again holds steady
o Initially, the quad heading is outside of 10 degree threshold so it corrects back using the
feedback via telemetry information.
 Interesting that there aren’t more data points in the West direction.
 The team liked these results better than the first one because of the more precise “box” pattern.
A-9
A.3.3.4
Box Test Number 3 - Crash into Divider
Notes:

Quadcopter started off fairly well, but ended up driving more NE than N. Eventually we simply
ran out of space and hit the divider wall before the turn had finished executing.
Results:
Summary:
 Quad was consistently below 0, but on the border of the threshold
o This caused the quad to slightly drift off course
 Quadcopter started pointing north-eastern. The momentum of this initial heading may have
resulted in the crash.
o We simply ran out of room to fly.
 However, you can see the quadcopter begin to execute its western turn when it makes contact
with the divider, from which point it came falling down.
A-10
A.3.3.5
Test Number 4: +/-10 Degree Heading Window
Notes:


The quadcopter was started much more northerly to help prevent the crash of the last test.
The team was very happy with these results.
Results:
Summary:
 Good data seen here - quad holds fairly straight, then consistently yaws towards destination
heading. Once at desired heading, again holds steady.
o The orientation feedback correction can easily be seen in the initial hold of 0 degrees as
well as the hold at -180 degrees at the end.
 The small spike in the middle could be a result of the quadcopter reaching -90 degrees, drifting
beyond the window, and correcting back to -90 before moving on to the -180 heading.
A-11
A.4
A.4.1
RC
Reactions for Following
Name: Robert Hoff, Spencer Olson, Jared Zoodsma
Date: 4/8/2014
Test Number: 1
Item under test: RC values being written by Arduino
Testing Parameters:
 Arduino program Follow_test_4
 Arduino
 GUI
 Pi code
 MultiWii_Telemetry_RevG.py
Notes:



Full communication test was run and quadcopter was plugged into open-sourced Multiwii GUI in
order to view RC inputs
Quadcopter was moved to various positions in relation to the base station in order to check
whether the RC values sent corresponded to the desired action of the quadcopter in order to get it
back to the base station
Example, putting the quadcopter to the north east of the base station would make the pitch go
down and the roll go to the left
Results:
 The RC values were recorded by screenshot along with associated latitude and longitude
differences
 RC values corresponded as expected for all positions
 Screenshots are shown in the screenshots folder and are named their respective latitude and
longitude differences
A-12
A.4.2
Reactions for Following with Flight
Name: Robert Hoff, Spencer Olson, Jared Zoodsma, Drew Brandsen
Date: 4/8/2014
Test Number: 2
Item under test: RC values being written by Arduino, and associated quadcopter flight
Testing Parameters:
 Arduino
 GUI
 Pi code
Notes:



All telemetry data was sent in the communication loop
code was modified to move the coordinates of the base station to ten meters north of the actual
position in order to provide safety of those near the base station
quadcopter was flown and the follow program was enabled, forcing the quadcopter to fly toward
the base station
Results:
 Initial flights yielded unpredictable flight patterns, but code was modified
 Quadcopter behaved as expected, flying toward the base station from any direction
 Wind was too unpredictable for stable flight, so full program could not be run with moving base
station
A-13
A.5
A.5.1
Sensors
Barometer
A.5.1.1
Positional Tests
 These tests are to determine best method for acquiring accurate barometer information
 The team is concerned that the prop wash will affect the barometer sensitivity. To counteract this,
the team is evaluating using foam and mesh to protect the barometer as well as a control (no
barometer protection)
 Initial “eyeball” test was done on 3/13/2014 and it appeared that no barometer cover resulted in
the best performance (this test was done with Drew, Spencer, and Jared)
A.5.1.1.1
Positional Test 1
Name: Jared Zoodsma
Date: 3/14/2014
Test Number: 1
Item under test: Ground Test No Prop Wash
Testing Parameters: (Ex. software version “xyz” used)
 Foam sponge, mesh sponge were used to cover the barometer as well as control (no cover)
 Quadcopter was placed on the ground without props running
 information was gathered for 1 minute
 tests were performed in the engineering building
 Telemetry.py (RevF) used on quadcopter
 An altitude of 0.0m was expected
Notes:
1. Possible disturbances may have come from people entering and leaving the building (this could
result in an overall pressure change)
2. All 3 options were tested with quadcopter on ground without props running
3. Used last 20 samples from 1 minute of data acquisition
Results:
Control
Foam
Mesh
Average
0.17
0.05
-0.12
Median
0.2
0.08
-0.23
Standard Dev.
0.17
0.16
0.33
A-14
A.5.1.1.2
Positional Test 2
Name: Jared Zoodsma
Date: 3/14/2014
Test Number: 2
Item under test: Table Test No Prop Wash
Testing Parameters: (Ex. software version “xyz” used)
 Foam sponge, mesh sponge were used to cover the barometer as well as control (no cover)
 Quadcopter was placed on a table (3.25’) without props running
 Information was gathered for 1 minute
 Tests were performed in the engineering building
 Telemetry.py (RevF) used on quadcopter
Notes:
 Possible disturbances may have come from people entering and leaving the building (this could
result in an overall pressure change)
 All 3 options were tested with quadcopter on ground without props running
 Used last 20 samples from 1 minute of data acquisition
 Communication was initialized on the ground (so it was set as 0) then moved to the table where
values were recorded
 An altitude of (1 meter = 3.25’) was expected
Results:
Control
Foam
Mesh
Average
0.98
1.27
0.37
Median
0.96
1.24
0.34
Standard Dev.
0.12
0.11
0.18
A-15
A.5.1.1.3
Positional Test 3
Name: Jared Zoodsma, Drew Brandsen
Date: 3/15/2014
Test Number: 3
Item under test: Ground Test with Prop Wash
Testing Parameters: (Ex. software version “xyz” used)
 Foam sponge, mesh sponge were used to cover the barometer as well as control (no cover)
 Quadcopter was placed on the ground with props running
 Information was gathered for 1 minute
 Tests were performed in the engineering building
 Telemetry.py (RevF) used on quadcopter
Notes:
 Possible disturbances may have come from people entering and leaving the building (this could
result in an overall pressure change)
 All 3 options were tested with quadcopter on ground with props running @ 50%
 Used last 50 samples from 1 minute of data acquisition
 An altitude of 0.0m was expected as the quadcopter was initialized on the ground.
Results:

Control
Foam
Mesh
Average
0.43
0.17
0.54
Median
0.43
0.16
0.59
Standard Dev.
0.16
0.17
0.27
Mesh option was deemed a poor option since it showed poor performance in all three evaluation
areas and was thrown out at this point
A-16
A.5.1.1.4
Positional Test 4
Name: Jared Zoodsma, Drew Brandsen
Date: 3/15/2014
Test Number: 4
Item under test: Table Test with Prop Wash
Testing Parameters: (Ex. software version “xyz” used)
 Foam sponge was used to cover the barometer as well as control (no cover)
 Quadcopter was placed on a table (3.25’) with props running
 Information was gathered for 1 minute
 Tests were performed in the engineering building
 Telemetry.py (RevF) used on quadcopter
Notes:
 Possible disturbances may have come from people entering and leaving the building (this could
result in an overall pressure change)
 Both options were tested with quadcopter on ground without props @ 50%
 Used last 50 samples from 1 minute of data acquisition
 Communication was initialized on the ground (so it was set as 0) then moved to the table and
props were initialized and values were recorded
 An altitude of (1 meter = 3.25’) was expected
Results:
Control
Foam
Average
1.11
1.02
Median
1.15
1.04
Standard Dev.
0.29
0.28
A-17
A.5.1.1.5
Summary of Positional Tests
Summary:
 Mesh cover was thrown out due to low performance in all metrics
Overall Averages

Control
Foam
Average
0.1725
0.1275
Median
0.1850
0.1300
Standard Dev.
0.1850
0.1800
At first glance, there seems to be very little difference in performance, if any between the control
(no cover) and the foam cover. However, after further inspection it appears that the foam had
better performances when prop wash was introduced (see table below)
Overall Averages (w/ Prop Wash)


Control
Foam
Average
0.270
0.095
Median
0.290
0.100
Standard Dev.
0.225
0.225
Initially standard deviation was weighed the highest of all metrics, however there seemed to be an
insignificant difference in standard deviation between control and foam.
When prop wash was introduced, foam resulted in better averages and medians than the control.
Therefore, Foam appears to be the best option for consistent and accurate altitude information
A-18
A.5.1.2
Flight Tests
Name: Jared Zoodsma, Drew Brandsen
Date: 3/15/2014
Test Number: 1
Item under test: Quadcopter Barometer - actual flying
Testing Parameters: (Ex. software version “xyz” used)
 Foam sponge was to cover the barometer as well as control (no cover)
 Quadcopter was initialized on the ground
 Information was gathered for 1 minute
 Tests were performed in the engineering building
 Telemetry.py (RevF) used on quadcopter
 An altitude of 2.0m was expected
Notes:


Possible disturbances may have come from people entering and leaving the building (this could
result in an overall pressure change)
Used last 50 samples (not the last 2 samples) from 1 minute of data acquisition
Results:

Control
Foam
Average
0.83
1.90
Median
0.78
1.94
Standard Dev.
0.45
0.31
Foam has shown to be the better option after the flight test. Using foam results in better standard
deviation and significantly better accuracy.
A-19
A.5.2
Compass
A.5.2.1
A.5.2.1.1
Compass Precision
Summary of Compass Precision Tests
Name: Drew Brandsen, Jared Zoodsma
Date: 4/13/2014
Item under test: MultiWii Compass
Testing Parameters:
 GUI
 Pi code
Notes:




Tests were done outside near the garage door of the engineering building
There was real-life disturbance such as wind
Sensor data was streamed wirelessly to the GUI to be recorded
Reported order: Latitude, Longitude, Altitude, Heading, Time
Conclusion:
 The EMF problem is solved as the motors no longer have an effect on the compass value.
 The compass data for tests 1, 2, 5, and 6 show an immediate drop to start the test. For some
reason, there was a 20-30 degree difference between a “0-heading” on the ground and a “0heading” in the air. As a result, the quadcopter immediately turned to correct for this change.
o It seems that the ground had some effect on the compass data which caused this problem.
 The quadcopter flew well with both barometer and heading modes activated. While the data
shows that the flight modes are working, there is nothing in software that the team can change to
improve the results (all onboard MultiWii software).
o To improve the results the team would have to look to improved hardware
A-20
A.5.2.1.2
Compass Test Number 1 - Centered at Zero, Takeoff with Mag Mode On
Notes:

The quadcopter was started facing a 0 degree heading before taking off. Mag Mode was always
activated.
Results:
Average
3.2
Max
17.0
Min
-21.0
Standard Deviation 5.7
Summary:
 Though the data varied a fair amount in either direction, the quadcopter always recovered to a
“near-zero” heading
 The team was encouraged by the results; however, we would have liked to see extreme data
points closer to 0
A-21
A.5.2.1.3
Compass Test Number 2 - Centered at Zero, Takeoff with Mag Mode On
Notes:

The quadcopter was started facing a 0 degree heading before taking off. Mag Mode was always
activated.
Results:
Average
3.8
Max
19.0
Min
-23.0
Standard Deviation 7.0
Summary:
 Though the data varied a fair amount in either direction, the quadcopter always recovered to a
“near-zero” heading
 The team was encouraged by the results; however, we would have liked to see extreme data
points closer to 0.
 This test had a higher standard deviation than the previous test.
A-22
A.5.2.1.4
Compass Test Number 3 - Fly to Zero Heading, then turn on mag mode
Notes:

The quadcopter took off and was “yaw-ed” to a 0 degree heading. At that point, mag mode was
activated until landing.
Results:
Average
-7.9
Max
18.0
Min
-33.0
Standard Deviation 9.2
Summary:
 It appears that Test 3 had worse results than the previous two tests. One likely cause of this is
that it was very hard to activate Mag Mode exactly when the heading was zero.
 This means that Mag Hold was likely activated around -5 degrees, which would yield a much
better average heading.
A-23
A.5.2.1.5
Test Number 4 - Fly to zero heading, then turn on mag mode
Notes:

The quadcopter took off and was “yaw-ed” to a 0 degree heading. At that point, mag mode was
activated until landing.
Results:
Average
2.2
Max
38.0
Min
-20.0
Standard Deviation 8.6
Summary:
 Again it appears that Test 4 had worse results than Tests 1 and 2. One likely cause of this is that
it was very hard to activate Mag Mode exactly when the heading was zero.
 This means that Mag Hold was likely activated around 3 degrees, which would yield a much
better average heading.
 This test also had one of the highest extreme values, but once again the quadcopter responded by
bringing the heading back to near-zero.
A-24
A.5.2.1.6
Compass Test Number 5
Notes:
 The quadcopter was started facing a 0 degree heading before taking off. Mag Mode was always
activated.
 Baro mode was activated at roughly 2 meters to see how well altitude was held while mag mode
was also being held
Results:
Average
2.6
Max
23.0
Min
-28.0
Standard Deviation 9.4
A-25
Average
2.1
Max
2.9
Min
0.1
Standard Deviation 0.4
Summary:
 Compass data was again similar to the previous 4 tests. There was a good average but again a
poor standard deviation
 Barometer data looks very good with a standard deviation of 0.4. There is the one outlier near the
end of the test that is bad.
o The problem with barometer is that it is only relative to air pressure (which can change
quickly), so what may look consistent in the data was actually much worse flight
performance relative to the ground.
A-26
A.5.2.1.7
Compass Test Number 6
Notes:


The quadcopter was started facing a 0 degree heading before taking off. Mag Mode was always
activated.
Baro mode was activated at roughly 2 meters to see how well altitude was held while mag mode
was also being held
Results:
Average
4.3
Max
43.0
Min
-52.0
Standard Deviation 12.6
A-27
Average
2.0
Max
3.0
Min
-0.1
Standard Deviation 0.7
Summary:
 Compass data was again similar to the previous 4 tests. There was a good average but again a
poor standard deviation. Only at the beginning was the data poor
 Barometer data looks good again with a standard deviation of 0.7. There are two outlier near the
middle of the test that is bad.
o The problem with barometer is that it is only relative to air pressure (which can change
quickly), so what may look consistent in the data was actually much worse flight
performance relative to the ground.
A-28
A.5.2.2
EMI Tests
The purpose of these tests is to determine the effectiveness of shielding the ESCs and how to
shield the flight controller.
A.5.2.2.1
EMI Test 1
Name: Drew Brandsen, Jared Zoodsma
Date: 3/17/2014
Test Number: 1
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 ESCs (not wires) wrapped in aluminum foil
 ESCs “dangling” (not taped close to internals of quadcopter)
 Test was done at table by wind tunnel
Notes:

Raspberry Pi was powered during these tests
Results:
 Compass was started at -100.0 degrees
 -103 degrees at full throttle
 -102 degrees at half throttle
 Compass was skewed 3 degrees when throttle was at full power
 Compass was skewed 2 degrees when throttle was at half power
A-29
A.5.2.2.2
EMI Test 2
Name: Drew Brandsen, Jared Zoodsma
Date: 3/17/2014
Test Number: 2
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 ESCs (not wires) wrapped in aluminum foil
 ESCs taped close to central part of quadcopter
 Test was done at table by wind tunnel
Notes:

Raspberry Pi was powered during these tests
Results:
 Compass was started at -100.0 degrees
 -102 degrees at full throttle
 -102 degrees at half throttle
 Compass was skewed 2 degrees when throttle was at full power
 Compass was skewed 2 degrees when throttle was at half power
Initial Summary:
 Aluminum foil wrapping significantly reduces compass EMI to an acceptable skew level (2
degrees)
Further Summary
 After a few test flights, it was noticed that the magnetometer slowly degraded to an error of
around 45 degrees (roughly the same as the control case).
 Additional foil was added and that seemed to correct the problem. But once again, test flights
proved the degradation of the magnetometer again.
A-30
A.5.2.2.3
EMI Test 3
Name: Drew Brandsen
Date: 3/17/2014
Test Number: 3
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 ESCs not wrapped
 ESCs dangling
 Test was done at table by wind tunnel
Notes:

Raspberry Pi was powered during these tests
Results:
 Compass was started at -180.0 degrees
 -163 degrees at full throttle
 -165 degrees at half throttle
 Compass was skewed 17 degrees when throttle was at full power
 Compass was skewed 15 degrees when throttle was at half power
A-31
A.5.2.2.4
EMI Test 4
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 4
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 mechanical connections loose
 Raspberry Pi case not on quadcopter
Notes:

This is a control test
Results:
 Initial position: -10 degrees
 Full throttle: -55 degrees
A-32
A.5.2.2.5
EMI Test 5
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 5
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 mechanical connections loose
 Raspberry Pi case is on quadcopter
Results:
 Initial position: -10 degrees
 Full throttle: -55 degrees
A-33
A.5.2.2.6
EMI Test 6
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 6
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 mechanical connections loose
 Raspberry Pi case not on quadcopter
Notes:

Flight Controller moved up ¾”
Results:
 initial position: -4 degrees
 full throttle: -14 degrees
A-34
A.5.2.2.7
EMI Test 7
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 7
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 Mechanical connections loose
 Raspberry Pi case not on quadcopter
Notes:


5.5X6.5” aluminum foil added directly below flight controller
o aluminum foil is grounded
o 4 screw holes located in the aluminum foil
Flight control sitting on quadcopter (no spacers added)
Results:
 Initial position: -7 degrees
 Full throttle: -48 degrees
A-35
A.5.2.2.8
EMI Test 8
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 8
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 Mechanical connections loose
 Raspberry Pi case not on quadcopter
Notes:


5.5X6.5” aluminum foil added directly below flight controller
o aluminum foil is grounded
Flight control sitting on quadcopter (no spacers added)
Results:
 Initial position: -16 degrees
 Full throttle: -56 degrees
A-36
A.5.2.2.9
EMI Test 9
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 9
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 Mechanical connections loose
 Raspberry Pi case not on quadcopter
Notes:


5.5X6” grounded steel plate placed directly below flight controller
Flight control sitting on quadcopter (no spacers added)
Results:
 Initial position: -165 degrees
 Full throttle: -169 degrees
A-37
A.5.2.2.10
EMI Test 10
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 10
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 Mechanical connections loose
 Raspberry Pi case not on quadcopter
Notes:


5.5X6” grounded thin steel sheet placed below flight controller
Flight control sitting on quadcopter (no spacers added)
Results:
 Initial position: -16 degrees
 Full throttle: -18 degrees
Summary:
 It appears adding distance and steel (not aluminum) helps prevent compass skew.
 Note: These tests only examined initial skew and did not investigate the change in skew over
time.
A-38
A.5.2.2.11
Effect of Power Supply Test 1
The purpose of these tests is to determine the source of compass skew.
Name: Drew Brandsen, Jared Zoodsma
Date: 3/17/2014
Test Number: 1
Item under test: Compass (Control Case)
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
Notes:

Raspberry Pi was powered during these tests
Results:
 Compass was started at -180 degrees
 -156 degrees at full throttle
 -160 degrees at half throttle
 Compass was skewed 34 degrees when throttle was at full power
 Compass was skewed 20 degrees when throttle was at half power
A-39
A.5.2.2.12
Effect of Power Supply Test 2
Name: Drew Brandsen, Jared Zoodsma
Date: 3/17/2014
Test Number: 2
Item under test: Compass
Testing Parameters:
 MultiWii is powered by a separate power supply
 Power distribution-sized ground plate is currently in place
Notes:

Raspberry Pi was powered during these tests
Results:
 Compass was started at -180 degrees
 -155 degrees at full throttle
 -160 degrees at half throttle
 Compass was skewed 35 degrees when throttle was at full power
 Compass was skewed 20 degrees when throttle was at half power
A-40
A.5.2.2.13
Skew over Time Test 1
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 1
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 Mechanical connections loose
Notes:


Raspberry Pi case placed under the flight controller
o Flight Controller is 1” above quadcopter plate
4 degrees will be recorded
o initial (no motors)
o immediately after motors are turned on
o 30 seconds after motors have been on
o 1 minute after motors have been on
Results:
 Initial position: -17 degrees
 Full throttle: -21 degrees
 After 30 seconds: -21 degrees
 After 1 minute: -21 degrees
A-41
A.5.2.2.14
Skew over Time Test 2
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 2
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 mechanical connections loose
Notes:

Flight Controller placed directly on a grounded steel plate
Results:
 Initial position: -24 degrees
 Full throttle: -27 degrees
 After 30 seconds: -27 degrees
 After 1 minute: -27 degrees
 -26 after 1:30
 -25 after 2:00
 -25 after 2:30
 -25 after 3:00
 -25 after 3:30
A-42
A.5.2.2.15
Skew over Time Test 3
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 3
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 mechanical connections loose
Notes:

Flight Controller placed directly on an ungrounded steel plate
Results:
 initial position: -23 degrees
 full throttle: -25 degrees
 after 30 seconds: -26 degrees
 after 1 minute: -27 degrees
 -27 degrees after 1:30
 -27 degrees after 2:00
 -27 degrees after 2:30
 -27 degrees after 3:00
 -27 degrees after 3:30
A-43
A.5.2.2.16
Skew over Time Test 4
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 4
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 mechanical connections loose
Notes:


Flight Controller placed directly on a grounded steel plate
After Throttle is established, it will be varied throughout the test
Results:
 initial position: -29 degrees
 full throttle: -31 degrees
 after 30 seconds: -32 degrees
 after 1 minute: -32 degrees
 -32 degrees after 1:30
A-44
A.5.2.2.17
Skew over Time Test 5
Name: Drew Brandsen, Jared Zoodsma
Date: 3/18/2014
Test Number: 5
Item under test: Compass
Testing Parameters:
 MultiWii is powered by the power distribution board
 Power distribution-sized ground plate is currently in place
 mechanical connections loose
Notes:


Flight Controller placed directly on an ungrounded steel plate
After Throttle is established, it will be varied throughout the test
Results:
 initial position: -31 degrees
 full throttle: -32 degrees
 after 30 seconds: -33 degrees
 after 1 minute: -33 degrees
 -33 degrees after 1:30
A-45
A.5.3
GPS
A.5.3.1
A.5.3.1.1
1 Hz Precision
Summary of 1 Hz GPS Precision
Name: Drew Brandsen, Jared Zoodsma
Date: 4/2/2014
Item under test: GPS accuracy
Testing Parameters:
 GUI code
 Adafruit GPS on Arduino (no external antenna)
 MediaTek GPS on Pi
Notes:


Arduino and Pi both placed in the same car (~1ft from each other)
No antennas used on either GPS modules
A-46
A.5.3.1.2
1 Hz Precision Test 1
Test Number: 1
Notes:


Car was stationary in the parking lot by the soccer fields
Bad Data: 3 erroneous distances points, 1 erroneous time point
Results:
Average abs(Lat):
0.2196
sd Lat:
0.3563
Average abs(Long):
0.6006
sd Long:
0.4884
Average Time:
0.9459
sd Time:
0.1961
Summary:
 We were pleased with these results since distances were within GPS margin of error
 Time was never more than 1.26 seconds apart; average was 0.96s\
 We believe that the few bad distance data points are coming from the Adafruit GPS sending
occasional blank data.
 We are unsure what is causing the few bad time data points seen throughout the tests.
A-47
A.5.3.1.3
1 Hz Precision Test 2
Test Number: 2
Notes:


Car was moving from North Parking lot to BB counter-clockwise (~15mph)
Bad Data: 11 erroneous distances points, 5 erroneous time points
Results:
Average abs(Lat):
3.2404
sd Lat:
3.5821
Average abs(Long):
4.2377
sd Long:
2.5401
Average Time:
0.8080
sd Time:
0.1666
Summary:
 The navigation path was along a curve, so we anticipated discrepancies along both lat and long.
 Time was still very consistent
 Location accuracy was not consistent, but neither was speed (varying from stop to 20 mph)
 Moving at 15mph (average), a 1 second delay could cause up to a 6.7 meters discrepancy.
o Therefore, we are pleased the results were, on average, within this threshold
o We wish that the gps data would have converged more at the end of the test when the
vehicle was stopped, however.
A-48
A.5.3.1.4
1 Hz Precision Test 3
Test Number: 3
Notes:


Car was stationary in BB parking lot
Bad Data: 2 erroneous distances points, 0 erroneous time points
Results:
Average abs(Lat):
2.5459
sd Lat:
0.0000
Average abs(Long):
3.1129
sd Long:
0.0000
Average Time:
0.7143
sd Time:
0.1145
Summary:
 Car was located under trees and by a dorm
o This could have resulted in poor signal strength
 GPS data was getting received (as indicated by changing time) but both GPS’s held their exact
location throughout the test.
A-49
A.5.3.1.5
1 Hz Precision Test 4
Test Number: 4
Notes:


Car moved from BB to north parking lot clockwise (~15 mph)
Bad Data: 4 erroneous distances points, 3 erroneous time points
Results:
Average abs(Lat):
4.0184
sd Lat:
4.0950
Average abs(Long):
4.1395
sd Long:
2.5752
Average Time:
0.7232
sd Time:
0.1219
Summary:
 Moving at 15mph, a 1 second delay could lead to 6.7meters discrepancy.
o Therefore, we are pleased the results were, on average, within this threshold
o However, there were several peaks well outside this range that could be improved.
A-50
A.5.3.1.6
1 Hz Precision Test 5
Test Number: 5
Notes:


Car drove Lake St to Burton on the E. Beltline with acceleration (max speed ~50mph)
Bad Data: 3 erroneous distances points, 1 erroneous time point
Results:
Average abs(Lat):
8.8128
sd Lat:
10.5910
Average abs(Long):
3.4960
sd Long:
1.5365
Average Time:
0.7327
sd Time:
0.1325
Summary:
 Moving at 45 mph, a 1 second delay could lead to 20.1meters discrepancy.
 The beltline is a North-South road, so only Latitude should see much change, which is roughly
what we see above.
 On average, the data was well within the possible discrepancy
o It’s possible that as the car went under the Calvin crossing that this is what gave us the
spike in data near time 80.
 The data also converged well at the end of the test.
A-51
A.5.3.1.7
1 Hz Precision Test 6
Test Number: 6
Notes:


Car was stationary in EB parking lot
Bad Data: 2 erroneous distances points, 1 erroneous time point
Results:
Average abs(Lat):
7.4368
sd Lat:
0.3078
Average abs(Long):
1.2452
sd Long:
0.0000
Average Time:
0.7737
sd Time:
0.1483
Summary:
 Time was very consistent and averaged 0.77 seconds.
 Latitude resulted in poor accuracy.
o However, we were located near the engineering building. This could have resulted in
poor signal fidelity.
A-52
A.5.3.2
A.5.3.2.1
10 Hz Precision
10 Hz Precision Summary
Name: Drew Brandsen, Jared Zoodsma
Date: 4/8/2014
Item under test: GPS Precision
Testing Parameters:
 Telemetry Code
 10 Hz GPS refresh rate
 GUI code running on Gateway
 Arduino code
Notes:



Arduino and Pi placed in same car (~6 inches apart)
Short GPS antenna used on base station
**Thanks to Spencer Olson for helping with post-test analysis in Excel
Conclusion:
 GPS precision is far improved with increased refresh rates.
 The randomly bad cases of Latitude data need to be investigated.
 The poor data around the BB parking lot may be due to the number of trees and buildings in that
area.
 It seems that precision increases with the speed that the car with the GPS modules is moving.
A-53
A.5.3.2.2
10Hz GPS Test Number 1 - Stationary at North Lot
Notes:
●
●
Sitting at north parking lot by track
started at 10:40
Results:
A-54
Time
Latitude
Longitude
Average
0.20
1.96
0.84
Standard Deviation
0.04
1.40
0.88
A-55
Data within 7.2 meter radius: 100%
Summary:
● We were pleased with the stationary results (100% within 6m radius)
Results from this test confirmed our suspicion that GPS locations were being reported in discrete
positional locations
A-56
A.5.3.2.3
Test Number 2 - Drive to BB Lot from North Lot 1
Notes:
●
●
●
●
Drive to BB Lot from North Lot 1
Test started at 10:45
Speed around 20 mph
Car stopped around 10:47:40
Results:
A-57
Time Latitude
Longitude
Average
0.2
4.52
5.80
Standard Deviation
0.05
3.55
2.26
A-58
Data within 7.2 meter radius: 60%
Summary:
●
●
Compared with the first test, GPS standard deviations increased and precision decreased
60% of data was within the expected area.
A-59
A.5.3.2.4
Test Number 3 - Stationary at BB Lot
Notes:
●
●
Stationary at BB parking lot
started at 10:49
Results:
A-60
Time Latitude
Longitude
Average
0.2
9.07
1.40
Standard Deviation
0.05
3.28
1.13
Data within 7.2 meter radius: 26%
A-61
Summary:
●
●
●
Compared with the North Lot stationary, this test was a whole lot worse.
This particular location seems to be very bad for GPS as other bad data was taken here on 4/2/14
Only 26% of the data was within the 6m radius, which is really poor.
A-62
A.5.3.2.5
Test Number 4 - Drive to North Lot from BB Lot 1
Notes:
●
●
●
●
Drive to North Lot from BB Lot 1
started at 10:51
Speed around 20 mph
stopped car at 10:52:45a
Results:
A-63
Time
Latitude
Longitude
Average
0.2
19.33
7.68
Standard Deviation
0.05
32.17
5.30
A-64
Data within 7.2 meter radius (All): 36%
Data within 7.2 meter Radius (21% omitted): 45%
Summary:
●
●
For an unknown reason, Latitude data became exorbitantly terrible. We are not sure if there was
a queuing issue within one of the GPS units that the comparison kept happening to an old position
even as the car moved away. Over time, the latitude data did correct itself, but not after being
over 100 meters off.
When this bad latitude data was taken off there was still only 45% of the data within the expected
range.
A-65
A.5.3.2.6
Test Number 5 - Drive to BB Lot from North Lot 2
Notes:
●
●
●
●
Drive to BB Lot from North Lot 2
started at 10:55
Speed around 20 mph
car stopped around 10:56:47
Results:
A-66
A-67
Time
Latitude
Longitude
Average
0.2
5.42
2.96
Standard Deviation
0.05
3.53
3.20
Data within 7.2 meter radius: 72%
Summary:
●
The second test to BB Lot from North Lot proved a little bit better with 72% of the data being
within range. However, this still isn’t great data.
A-68
A.5.3.2.7
Test Number 6 - Drive to North Lot from BB Lot 2
Notes:
●
●
●
●
Drive to North Lot from BB Lot 2
Speed around 20 mph
started at 10:59
stopped car at 11:00:40
Results:
A-69
A-70
Time
Latitude
Longitude
Average
0.2
6.07
2.76
Standard
Deviation
0.05
3.70
1.54
Data within 7.2 meter radius: 66%
Summary:
●
●
Just like the previous test, the second round of BB Lot to North Lot proved more precise with
66% of the data being within the expected range.
Nevertheless, this data is still isn’t great.
A-71
A.5.3.2.8
Test Number 7 - Drive Beltline North
Notes:
●
●
●
●
●
●
●
Drive on Beltline North
Hard acceleration to 45 mph
Average speed around 45 mph
stoplight at: 11:05:53
green 11:06:13
stoplight at: 11:08:06a
green 11:08:15a
Results:
A-72
Time
Latitude
Longitude
Average
0.2
6.27
3.95
Standard Deviation
0.04
4.51
2.20
A-73
Data within 7.2 meter radius: 56%
Summary:
●
●
The data gathered was actually quite good until the end when the Latitude data continued to
plummet farther and farther away. Possibly another queuing issue of data?
56% of the data landed within the expected radius.
A-74
A.5.3.2.9
Test Number 8 - i96 Eastbound
Notes:
●
●
●
●
Drive on i-96 eastbound
started at 11:11
Up to speed at 11:12:30
Average Speed around: 70mph
Results:
A-75
Time
Latitude
Longitude
Average
0.2
3.38
3.83
Standard Deviation
0.05
2.14
2.83
A-76
Data within 7.2meter radius: 81%
Summary:

This test gathered very good data with 81% of the data points landing within the expected range.
It seems that with higher speed (20 mph -> 45 mph -> 70 mph) that the GPS precision also
increased.
A-77
A.5.3.2.10
Test Number 9 - 28th W → E Paris N → Burton W
Notes:
●
●
●
●
●
●
●
Drive 28th W, E Paris N, Burton W: many stoplights, real life
started at 11:17
Top speed around 45 mph
normal speed around 38mph
Turn on East Paris north 11:20:15
full stop at light 11:21:54
Turn on Burton west 11:22:05
Results:
A-78
Time
Latitude
Longitude
Average
0.2
3.68
1.84
Standard Deviation
0.05
3.01
1.73
A-79
Data within 7.2 meter radius: 81%
Summary:
●
●
It is interesting that with this most real-life situation, the data gathered was far better than that
taken on Calvin’s Campus. The car traveled varying speeds (from stops at lights to 45 mph) and
also various directions.
81% of the data landed within the expected radius.
A-80
A.6
A.6.1
A.6.1.1
Video
RTSP Latency
RTSP Latency Test 1
Name: Spencer Olson, Jared Zoodsma
Date: 2/10/2014
Test Number: 1
Item under test: Video Latency
Testing Parameters:
 GUI Version: “TCP->GUI->Arduino->GUI_RevA”
 Gateway used to run GUI
 Video Parameters: 480 x 852, 24 frames per second
 RTSP used for video protocol
Notes:
Results: 1.8 second latency
A-81
A.6.1.2
RTSP Latency Test 2
Name: Spencer Olson, Jared Zoodsma
Date: 2/10/2014
Test Number: 2
Item under test: Video Latency
Testing Parameters:
 GUI Version: “TCP->GUI->Arduino->GUI_RevA”
 Gateway used to run GUI
 Video Parameters: 480 x 852, 10 frames per second
 RTSP used for video protocol
Notes: Video quality seemed very inconsistent and began to skip
Results: 2.3 second latency
A-82
A.6.1.3
RTSP Latency Test 3
Name: Spencer Olson, Jared Zoodsma
Date: 2/10/2014
Test Number: 3
Item under test: Video Latency
Testing Parameters:
 GUI Version: “TCP->GUI->Arduino->GUI_RevA”
 Gateway used to run GUI
 Video Parameters: 1080x1920, 24 frames per second
 RTSP used for video protocol
Notes: GUI struggled rendering full HD picture
Results: 2 second latency
A-83
A.6.1.4
RTSP Latency Test 4
Name: Spencer Olson, Jared Zoodsma
Date: 2/10/2014
Test Number: 4
Item under test: Video Latency
Testing Parameters:
 GUI Version: “TCP->GUI->Arduino->GUI_RevA”
 Gateway used to run GUI
 Video Parameters: 1080x1920, 10 frames per second
 RTSP used for video protocol
Notes: GUI struggled rendering full HD picture
Results: 2.4 second latency
A-84
A.6.2
Summary
Name: Drew Brandsen, Spencer Olson, Jared Zoodsma
Date: 4/18/2014
Item under test: Video Latency and Video Quality
Testing Parameters:
 Telemetry data was sent while video streaming
 Quad was flying (RC present) while video streaming.
Notes:



Weather was favorable with little wind.
Flights lasted around 2 minutes
Tests were done outside the Engineering Building garage door.
Conclusion:
 Although GStreamer produced more video jitter than the other protocols used, it proved to be the
best protocol due to significantly lower video latency
 The GStreamer video jitter was reduced by using higher frame rates
 After preliminary tests, GStreamer with 480p resolution and 60 fps provided the best results.
o Further tests can be performed to optimize video stream specifications.
o Graphs of the results are seen on the next page.
Graphs:
Results of Video Stream Protocol:
A-85
Results of Video Stream Specifications
A-86
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement