Thesis
NXTSim: A 2D simulator for a LEGO
Mindstorms NXT robot controlled by
MATLAB
BSc Thesis
Müttalip Küçük, 2524209
Department of Computer Science,
VU University Amsterdam, Netherlands
Supervisor:
Second reader:
Natalia Silvis-Cividjian
Anton Eliens
2015
Table of contents
Abstract..……………………………………………………………………………………………...2
Chapter 1. Introduction.…………………………………………………………………………….3
Chapter 2. LEGO Mindstorms robot controlled by MATLAB – Problem statement.…….4
Chapter 3. 2D simulators for robot programming – State of the art.……………………….7
3.1. Methods…………………………………………………………………………………..7
3.2. Results & discussion……………………………………………………………………8
Chapter 4. A 2D simulator – NXTSim..…………………………………………….……………11
4.1. Approach…………..……………………………………………………………………11
4.2. Sensors, actuators and environment.…..……………………………………………11
4.3. Parsing the code...……………………………………..………………………………13
4.4. User interface…………………………………………..………………………………14
4.5. Testing……….…………………………………………..……………………………..16
4.5.1. Line follower…………………………………………………………………16
4.5.2. Obstacle avoidance…………………………………………………………18
Chapter 5. Conclusion.……………………………………………………………………………19
References...………………………………………………………...………………………………20
Appendix..………………………………………………………………………...…………………21
User Manual…………………………………………………………………………………21
Abstract
The role of simulators in robotics research is important because of its efficiency, safety and
robustness of new algorithms. Research shows that a simulator for LEGO Mindstorms NXT
robots, which accepts MATLAB-code as input, does not exist. We developed a simulator to
solve this problem. In this thesis, we contribute a simulator for the LEGO Mindstorms NXT
robotic kit that accepts MATLAB-code as input. The main focus is on simplicity and
usefulness, because it will be used by students with no experience in programming.
2
Chapter 1. Introduction
In 1988, Mark Weiser introduced the term “ubiquitous computing” which means that
computing should be accessible everywhere and anywhere. In 1991, Weiser described his
idea as follows: “Specialized elements of hardware and software, connected by wires, radio
waves and infrared, will be so ubiquitous that no one will notice their presence” (Weiser,
1991). This vision involves advantages in business, public and private fields. Friedewald
(2011) describes applications of ubiquitous computing in retail, industrial production and
material management, transport logistics, personal identification and authentication, health
care and mobility. This shows the importance of computers in today’s world. Other
experiments predict that their role will grow further (Anderson et al., 2002). In 2002, there
were more than 150 million hosts connected to the Internet, and the number was growing
exponentially. Staranowicz et al. (2011) says that the need of robots is motivated by modern
applications in which robots are required to autonomously operate in close interaction with
humans. Hence it can be concluded that a future world without computers is not imaginable.
As a result of this growth, several branches within the computer science became interesting.
Examples of these branches are software engineering or artificial intelligence. Because of
the growth of the Internet during the years, a lot of knowledge can be reached online. This
makes it easy to learn programming. Nowadays, in the early years of school, basic
programming courses are given to students due the importance of IT in business. Because
there are a lot of ways to teach the basics of programming, the school has to choose an
environment, which is suitable for the school as well as for the students. The school wants an
environment, which has a lot of tools available. Students prefer an environment, which shows
quick results of their programming efforts. To satisfy both, robots can be used to teach
programming and became familiar with artificial intelligence (Kumar, 2004). A bad thing
about using robots to teach programming is that it demands a lot of time and effort from the
students and the instructor.
A widely used platform for teaching artificial intelligence and the basics of programming is the
LEGO Mindstorms NXT robotic kit. This robot is used by a lot of universities and there are
even competitions to compare the robots programmed by the students (Patterson-McNeill,
2001). As earlier said, using robots in program courses takes a lot of time. A reason for this
is that the duration of a class is limited and students need time to test their program with a
robot. A solution for this time problem is to use simulators more. This allows students to work
at home in their own time.
In this thesis, we present a program called “NXTSim”, which simulates a LEGO Mindstorms
NXT 2.0 robot in 2D. This simulation accepts a syntactically correct MATLAB code as input
for the robot. We intend to use it for the course “Pervasive Computing” lab sessions at the
VU University Amsterdam. The students of this course are beginners in MATLAB as well as
in using the robot.
This thesis is built as follows. In chapter 2, we describe the problem statement. In chapter 3,
we describe a systematic literature review about simulators and robots in general. Also, the
used methods, some results and discussion are given in this chapter. In chapter 4, we
describe the implementation of our 2D simulator in more technical detail, and test NXTSim
with two examples. In chapter 5, we end up with a conclusion and give some ideas for future
work. In the appendix, you can find a user manual for NXTSim.
3
Chapter 2. LEGO Mindstorms robot controlled
by MATLAB - Problem statement
The purpose of this project is to develop a 2D simulator for a LEGO Mindstorms NXT 2.0
robotic kit controlled by MATLAB. It will be used during the course “Pervasive Computing” at
the VU University Amsterdam. A requirement is that the performance speed has to be
acceptable. This means that it may not work too slowly. This chapter explains the
fundamentals of the hardware and the software of the project.
The LEGO Mindstorms robotic kit is dividable in three parts. The first part is the controller.
The controller shown in Figure 2 is the brain of the robot and where the program runs. The
controller has totally eight ports. Four of these port are the output ports located at the top of
the controller, and four are the input port located at the bottom. The four output ports at the
top are labelled as “A”, “B”, “C” and “USB”. The ports “A”, “B” and “C” should be connected
with actuators which will be explained later. The port “USB” is to make an USB-connection
with another device. The USB-port will not be used in this project. The four input ports at the
bottom are numbered from 1 to 4. These four ports should be connected with sensors which
will also be explained later. In short, the controller has four input ports (1, 2, 3 and 4) and
three output ports (A, B, and C).
Sensors are connected to the controller. There are a lot of sensors available for the LEGO
Mindstorms robotic, but we will use two of them. The first is the light sensor shown in Figure
3. This sensor measures the reflected light and returns a greyscale value between 0 (white)
and 800 (black). It is often used for line following. The second sensor is the ultrasonic sensor
shown in Figure 4. This sensor generates sound waves, and reads their echoes to detect
and to measure the distance to objects. The maximal range of this sensor is 255 cm. So the
returned value is between 0 and 255.
Figure 2. LEGO Mindstorms Figure 3. LEGO Mindstorms Figure 4. LEGO Mindstorms
controller.
light sensor.
ultrasonic sensor.
Actuators are also connected to the controller. There are also numerous actuators for the
LEGO Mindstorms robotic, but we would limit it to two actuators: motors and LEDs. The
motor shown in Figure 5 makes rotations for a given speed between 0 and 100. It can be
connected to a wheel. We used this actuator in this way. The second actuator is the LED
shown in Figure 6 which can be switched on or off. The colour can be red, yellow or green.
4
Figure 5. Motor.
Figure 6. LED
Figure 7(a) shows the robot which is used during the course “Pervasive Computing” and
which we will implement in the simulator. The robot consists of a controller, two light sensors,
one ultrasonic sensor and two motors. The other version of the robot is connected with LEDs
instead of motors, and will also be implemented. The environment of the robot is a white
surface on the floor where the robot would be placed on (see Figure 7(b)). This surface is 90
centimetres by 120 centimetres. There can be objects on this surface, so that we can
analyse the behaviour of the robot. The available objects are black lines, horizontal and
vertical walls and other 3D obstacles.
Figure 7. (a) LEGO Mindstorms robotic kit used during “Pervasive Computing”. (b) The
environment with the robot.
5
The software which will be written by the students to control the robot is in MATLAB. This
software uses the RWTH MATLAB toolboxes for LEGO (Aach et al., 2008). MATLAB is a
numerical computing environment which has a lot of tools and functions. This program is
widely used by schools and universities, because it is easy to learn to work with. We used
MATLAB 8 version R2012b during this project. The written MATLAB program will also be one
of the inputs of the simulator. Table 1 shows the commands that have to be recognized by
the simulator. Combinations of code will also be possible and valid. The first column shows
the code, the second column shows which part of the robot that code uses, and the third
column shows a short description.
Command
COM_CloseNXT('all')
close all
h = COM_OpenNXT()
COM_SetDefaultNXT(h)
COM_CloseNXT(COM_GetDefau
ltNXT())
CloseSensor(SENSOR_1)
OpenLight(SENSOR_1, ‘active’)
GetLight(SENSOR_1))
Category
Initialization
Initialization
Initialization
Initialization
Initialization
Description
Initialize the connections of the robot.
Sensor
Light sensor
Light sensor
OpenUltrasonic(SENSOR_1)
m = NXTMotor('AC', 'Power', 50)
Ultrasonic
sensor
Ultrasonic
sensor
Motor
m.Power = 10
m.SendToNXT()
Motor
Motor
m.Stop (‘off’)
m.Stop(‘brake’)
SwitchLamp(Motor_A, ‘on’)
SwitchLamp(Motor_A, ‘off’)
x = 10
x > 100
vec(i) = 10
display(x)
If (true) … else if (true) … end
Pause (0.5)
% this is a comment
While (true) … end
For i = 1:20 … end
Break
Motor
Motor
LED
LED
General
General
General
General
General
General
General
Loop
Loop
Loop
Close sensor connected in port 1.
Open light sensor connected in port 1.
Return value of light sensor connected
in port 1.
Open ultrasonic sensor connected in
port 1.
Return value of ultrasonic sensor
connected in port 1.
Set the speed of motors connected in
port A and C to 50, is assigned to
variable m.
Assign the power of variable m to 10.
Send the instructions assigned to m to
the controller.
Turn the motors in m off.
Set speed of the motor in m to 0.
Switch LED connected in port A to on.
Switch LED connected in port A to off.
Assign 10 to variable x.
Return a Boolean expression.
Assign 10 to place i of vector.
Print variable x.
If- and else if- statement.
Pause for 0.5 milliseconds.
Comment for the user.
While-loop.
For-loop.
Break out (for or while) loop.
GetUltrasonic (SENSOR_1)
Close the connection.
Table 1. Typical MATLAB commands used for our simulator.
6
Chapter 3. 2D simulators for robot
programming - State of the art
3.1. Methods
A systematic literature review provides the relevant articles from all available literature to
answer the research question or to learn more about other interests. To perform this review,
we followed the steps defined by Kitchenham (2004) and Khan et al. (2001). The following
stages and activities are performed:
Stage 1: Planning the review
Activity 1.1: Identification of the need for a review
Activity 1.2: Development of a review protocol
Stage 2: Conducting the review
Activity 2.1: Identification of research
Activity 2.2: Selection of primary studies
Activity 2.3: Study quality assessment
Activity 2.4: Data extraction and monitoring
Activity 2.5: Data synthesis
Stage 3: Reporting the review
Activity 3.1: Communicating the results
Firstly, we performed a search to check whether there are systematic reviews about
programming simulators. No results were found. Further, we examined the state of research
in programming simulators and we used the following research questions:
RQ1: What is the platform of the robot?
RQ2: Which languages are used for developing a simulator?
RQ3: What are the components of the simulator from its maker point of view?
RQ4: Which objects are defined in the simulator from its user point of view?
In April 2015, a systematic review was undertaken in the following online digital libraries: (a)
ACM Digital Library, (b) ScienceDirect and (c) Google Scholar. Only results written in English
were selected.
The search string used was: (simulation OR simulator OR simulate OR 2D or 3D) AND (robot
OR robotic OR robotics OR LEGO OR Mindstorms) AND (programming OR program OR
software OR JAVA) AND (teaching OR teach OR learning OR learn OR educate OR
education OR educational OR school). Table 2 shows the protocol executed for each
database.
Three criteria for exclusion (EC) articles were also identified:
EC1: The article is not about robots used in programming or artificial intelligence courses.
EC2: The article is about a robot instead of a simulator.
EC3: The simulator presented in the article is not achievable due to the duration of project
and the small development team.
The initial search yielded 226 papers. Firstly, we selected on titles and abstracts using the
inclusion criteria. Generally, it was easy to select on titles and abstracts, but there were
articles of which we had the read the full text to decide whether it does not fit the exclusion
criteria. Table 3 shows more detail about the selection we made.
7
Database
ACM Digital
Library
Protocol
(simulation OR simulator OR
simulate OR 2D or 3D) AND (robot
OR robotic OR robotics OR LEGO
OR Mindstorms) AND (programming
OR program OR software OR JAVA)
ScienceDirect
Note
The search string is shortened,
because it returned few results.
The following text is deleted:
(teaching OR teach OR learning
OR learn OR educate OR
education OR educational OR
school)
Open access articles
(simulation OR simulator OR
simulate OR 2D or 3D) AND (robot
OR robotic OR robotics OR LEGO
OR Mindstorms) AND (programming
OR program OR software OR JAVA)
AND (teaching OR teach OR learning
OR learn OR educate OR education
OR educational OR school)
Google Scholar (simulation OR simulator OR
simulate OR 2D or 3D) AND (robot
OR robotic OR robotics OR LEGO
OR Mindstorms) AND (programming
OR program OR software OR JAVA)
AND (teaching OR teach OR learning
OR learn OR educate OR education
OR educational OR school)
Table 2. The specific protocol executed in each database.
Database
ACM Digital
Library
ScienceDirect
Google
Scholar
Total
Articles resulting from the search
26
EC1
16
EC2
2
EC3
6
Selected
2
100
100
51
55
9
13
36
24
4
8
122
24
Table 3. Summary of article selection.
66
14
226
Table 3 shows that around 55% of the articles resulting from the search were excluded,
because these are covered by EC1. This means that the resulting article is about another
subject in the science and not about robots used in programming or artificial intelligence
courses. Over 10% of the resulting articles were excluded due EC2, because these articles
focus more on comparison of several robots and not or less on the simulation which uses a
specific robot. Approximately 30% of the articles resulting from the search were not accepted
due EC3, because the presented simulation is not achievable due to the duration of the
project and the small development team.
3.2. Results & discussion
Table 4 shows an overview of all relevant articles resulting from the literature research. For
each article, Table 4 shows the following attributes: (a) column 1: the name of the article or
the presented simulator, (b) column 2: Robot type shows the type of the used robot, (c)
Language simulator shows the language which is used by the simulator, (d) Language input
file shows the language which is used as input by the simulator, (e) Components presents
the component of the simulator, (f) Defined objects present which objects are defined in the
simulator.
8
Article or
simulator
Tian (2007)
Gonçalves
et al. (2009)
Robot
platform
LEGO
Mindstorms
RCX
LEGO
Mindstorms
NXT
Language
simulator
ODE, C++
(OpenGL),
ROBOLAB
SimTwo,
ODE, C++
(OpenGL),
XML
Robotran
Language
input file
ROBOLAB
leJOSl
Defined objects
GUI, Object
Environment,
Properties
Object
Environment
Predefined robots and
environments
IDE for editing
code, simulator
for visualizing
robot
GUI, 3D,
Documentation,
Tutorial
Robot, line, circle, square,
move direction
Robot with two motors
and light sensor, line
Meyer et al.
(2009)
LEGO
Mindstorms
RCX
Webots
C, C++,
Java,
Matlab,
Python
C++
(OpenGL)
leJOS
None
GUI, 3D,
Documentation
Multiple robots
Rob. Tbx.
Many, also
LEGO
Mindstorms
NXT
Robotic
platforms in
manufacturing
plants
Not specified
C-like
MATLAB
Robots with many sensors
Player/Stage
Not specified
Any
language
Gazebo
Not specified
ROS
Not specified
Simbad
Not specified
CARMEN
Not specified
C, C++,
Python,
Java
C, C++,
Python,
Java
C++,
Python,
Octave,
LISP,
Java, Lau
Java,
Python
C, Java
Sarsi
Not specified
C, C++,
Java
Self-made
software
GameBot
interface
GUI, 2D/3D,
Documentation,
Tutorial
2D,
Documentation,
Tutorial
3D,
Documentation,
Tutorial
2D/3D,
Documentation,
Tutorial,
Community of
researchers
3D,
Documentation,
2D,
Documentation,
GUI, 3D,
Documentation,
MRDS
Not specified
Visual
programming
language
GUI, 3D,
Documentation,
Tutorial
MissionLab
Not specified
VPL, C#,
Visual
Basic,
Jscript,
IronPython
VPL
Self-made
software
GUI, 3D,
Documentation,
Easy-Rob
leJOS
Components
Any
language
Not specified
Not specified
Table 4. Context of the articles.
9
Customizable robots,
obstacles, walls, line,
more
Robot with many sensors,
obstacles and accessories
Robot with many sensors,
obstacles and accessories
Robots, nodes
Single and multiples
robots, obstacles
Real mobile-robot
platforms and sensors
Several default maps,
allows creating own maps,
a variety of sensors
A variety of robotic
platforms and sensors
Multi-agent robotic
mission specification,
several tools to plan the
missions
There is a relation between the attributes Language simulator and Language input file shown
in Table 4. Figure 8 shows the way the two compilers translate the language of the input
code to the language of the simulator. Firstly, the language of the input file is compiled to an
intermediate code which is the bridge between the languages of the input file and the
simulator. Then, the intermediate code is compiled to the language of the simulator which will
show a simulation that interprets the input file.
Language input file Compiler (to intermediate code) Intermediate code Compiler (to language simulator) Language simulator Figure 8. Transformation sequence.
We analyse the results of the systematic review and try to answer the research questions
defined in paragraph 3.1.
RQ1: What is the platform of the robot?
In most of the retrieved articles, it is not specified for which robot type it is be used (Webots,
Easy-Rob, Rob. TX., Player/Stage, Gazebo, ROS, Simbad, CARMEN, USARSim, MRDS,
MissionLab). So this allows it to use it for different robot types. The other articles presents a
simulator for the RCX (Tian, 2007; Meyer et al., 2009) and NXT (Gonçalves et al, 2009)
models of Lego Mindstorms.
RQ2: Which languages are used for developing a simulator?
The results show that two of the most used programming languages are C++ and Java
(Webots, Easy-Rob, Player/Stage, Gazebo, ROS, Simbad, USARSim). Other often used
programming languages are C and Python (CARMEN). The following programming
languages are used by a few articles: VPL (MissionLab), LISP, Lua, Visual Basic, Jscript,
IronPython (MRDS), Octave, ODE, Robotran (Meyer et al., 2009), ROBOLAB (Tian, 2007),
SimTwo (Gonçalves et al., 2009), XML and a C-like language (Rob. Tbx.).
RQ3: What are the components of the simulator from its maker point of view?
Almost all articles contain a graphical user interface (GUI) and documentation. Most of the
GUI’s are in 3D (Webots, Easy-Rob, Gazebo, Simbad, USARSim, MRDS, MissionLab), a
few are in 2D (Player/Stage, CARMEN), and a few in both 2D and 3D (Rob. Tbx., ROS).
Another well common component in the articles is a tutorial. Other mentioned components
are object environments (Gonçalves et al., 2009), properties (Tian, 2007), and an IDE for
editing the code (Meyer et al., 2009). One article uses MATLAB as the language of the input
file (Rob. Tbx., ROS).
RQ4: Which objects are defined in the simulator from its user point of view?
In most of the articles, the user is able to customize the robot with several sensors
(Gonçalves et al., 2009; Webots, Easy-Rob, Rob. Tbx., Player/Stage, Gazebo, ROS,
Simbad, CARMEN, MRDS, MissionLab). All simulators focus on the programming language
and the graphical user interface (Tian, 2007; Meyer et al., 2009; USARSim). So the defined
objects are clearly the same: robots, obstacles, lines and nodes.
10
Chapter 4. A 2D simulator for LEGO
Mindstorms controlled by MATLAB - NXTSim
This chapter is about the implementation of the hardware which is mentioned in the previous
chapter. We will also tell about the parsing of the MATLAB code to intermediate code. At the
end of this chapter, we present the user interface and the design of NXTSim.
4.1. Approach
The main requirement is as follows: Given a MATLAB code as input file, develop a 2D
program which simulates the physical LEGO Mindstorms NXT robot. Before developing the
simulator we had to choose which programming language we would use. Our team1 is most
familiar with JAVA, so we decided that the language of the simulator is JAVA. Then, we
thought about ideas to develop a digital version of the hardware and the environment
mentioned in the previous chapter, and about parsing the software.
4.2. Sensors, actuators and environment
Before explaining the implementations, we have to notice that 10 cm in the physical world is
64 pixels in the simulation. Firstly, we design a 68 pixel by 119 pixels image shown in Figure
9 which represents the robots which we will use. The left one is connected with motors, and
the right one with LEDs. The implementations of motors and LEDs will be explained later.
Then, we thought about the implementations of the three parts of the hardware which are
controller, sensors and actuators. We do not have to implement the controller, because that
is the program itself. The first sensor to implement is the light sensor. The light sensor shown
in Figure 10 has a width of 12 pixels and a height of 18 pixels. We select nine coordinates of
each light sensor shown as yellow in Figure 10. These nine coordinates will measure the
colour under these coordinates. After knowing the nine colours from each sensor, the
majority of the nine colours will be the colour of that sensor. The nine coordinates are
combinations of the x-coordinates 1, 6 and 12, and the y-coordinates 1, 9 and 18. The
second sensor is the ultrasonic sensor which is implemented as follows. We select five
coordinates with x-coordinates 1, 17, 34, 51 and 68 and with y-coordinate 92 shown in Figure
11. From these five points, we go one pixel further till one of these point measures an
obstacle. If we measure an obstacle, we save the number of the pixel we went through.
Otherwise, if we do not measure an obstacle and the maximum range of the ultrasonic
sensor is reached, then we saved the maximum range as observed value to an obstacle.
Figure 9. The two robots
in the simulation.
Figure 10. Nine coordinates Figure 11. Five coordinates
of the light sensor.
of the ultrasonic sensor.
1
The team consists of one student which is Müttalip Küçük.
11
The first actuator implemented is the motor. We do not have to implement a graphical item
for the motor. The only important thing about the motor is that the simulation has to know
whether it is on or off, and the speed of the motor. Given these two data for both motors, we
move the robot in the simulation in two steps. The first step is to move the robot with the
lowest speed of the two motors in the direction which is saved in the program. The second
step of moving the robot is to calculate the difference between the speed of the two motors,
and rotate the robot. This looks like the robot drives a curve. When the two motors have
equal speed, then the degree of rotation is 0 which will not cause a curve. We will explain
this with the example (see Figure 12) to make the idea clear. For example, if the speed of the
left motor is 30 and the right motor is 10, and both motors are on, then the robot has to drive
to the right with a certain angle. The first step of our implementation tells that the right motor
has the lowest speed which is 10. The robot drives straight forward with speed 10 which is
shown as two red arrows in Figure 12(b). The second step calculates the difference of the
two speeds which is 20. The following formula calculates the degree of a rotation:
rotation = tan-1 ( (speed left motor – speed right motor) / width robot ),
where width robot = 68.
Using this formula, we calculate the angle A shown in Figure 12(c). Then, we rotate the robot
with this angle shown in Figure 12(d). Finally, the robot has moved. Figure 12(a) and 12(e)
show respectively the robot before and after it was moved.
Figure 12. (a) Before the robot moved, (b) before step 1 (c) after step 1 and before step 2,
(d) after step 2, (e) after the robot moved.
The implementation of the LED is the simplest, because the program just had to select one
of the robots in Figure 13. The robot can be connected with at most three LEDs. If the LED is
connected and is switched to on, the LED becomes green. Otherwise the LED will be red.
Figure 13. All possible states of a robot connected with at least one LED.
12
The environment in the physical world is a 90 cm by 120 cm white surface. We designed a
grid with nine columns and twelve rows. Each cell represents a part which is 10 cm by 10
cm, but in the program it is 64 pixels by 64 pixels.
4.3. Parsing the code
Our assumption is that the MATLAB file is syntactically correct. The first step is to parse the
MATLAB file to an intermediate code, see the left rectangle in Figure 14. In this step, we
determine which category a MATLAB command belongs to. Table 5 shows all the possible
categories with the corresponding MATLAB commands. The second step is to parse the
intermediate code to the language of the simulator, which are both in JAVA. The code in the
language of the simulator performs our implementations of the simulator.
Language input file Compiler (to intermediate code) Intermediate code Compiler (to language simulator) Language simulator Figure 14. Transformation sequence.
Category
Space
Initialization
For
While
If
Assignment
SendToNXT
Stop
Break
Pause
SwitchLamp
OpenLight
GetLight
OpenUltrasonic
GetUltrasonic
CloseSensor
Display
Expression
Unknown
Command(s)
All commands consisting of whitespaces or tabs.
COM_CloseNXT('all')
close all
h = COM_OpenNXT()
COM_SetDefaultNXT(h)
COM_CLoseNXT(COM_GetDefaultNXT())
For i = 1:20 … end
While (true) … end
If (true) … else if (true) … end
m = NXTMotor('AC', 'Power', 50)
m.Power = 10
x = 10
vec(i) = 10
m.SendToNXT()
m.Stop (‘off’)
m.Stop(‘brake’)
Break
Pause (0.5)
SwitchLamp(Motor_A, ‘on’)
SwitchLamp(Motor_A, ‘off’)
OpenLight(SENSOR_1, ‘active’)
GetLight(SENSOR_1))
OpenUltrasonic(SENSOR_1)
GetUltrasonic (SENSOR_1)
CloseSensor(SENSOR_1)
display(x)
x > 100
All command which do not belong to one of the categories above.
Table 5. Categories in parsing a MATLAB command.
13
4.4. User interface
We decided to make two screens: one screen for initialization of the ports, and another
screen to input the file and to run the simulation.
Figure 15 shows the initialization screen. The main goal of the initialization screen is to
connect sensors with ports 1, 2, 3 and 4 and actuators with ports A, B and C. This screen is
divided in two parts. The left part shows the controller with the connected sensors and
actuators. The right part is to select a sensor or actuator, or to end the communication
process. All the seven ports are clickable. After a click on one of the ports, the right part of
the screen will show one of the following two scenarios: a menu with the available sensors,
or a menu with the available actuators. When the initialization is completed, the user clicks
on the OK button.
Figure 15. The initialization screen.
The second screen of our program is the simulation screen shown in Figure 16. The goal of
this screen is to build the environment of the robot, to read in the MATLAB file, and to run the
simulation. The simulation screen contains a menu bar at the top, a tools panel at the left, an
environment panel at the centre, a code panel at the top right, and an output panel at the
bottom right.
The menu bar at the top contains the options “File” and “Run”. The “File”-option can be used
to clear the environment, to import the file with the MATLAB code, or to quit the program.
The user can delete all objects in the built environment by File>Clear environment. After
clicking on File>Import MATLAB-file, the user can browse through his or her folders to select
the MATLAB-file. To quit the program, the user selects File>Quit. The option Run>Run is to
14
run the simulation and becomes clickable after the robot is placed and the MATLAB-file is
imported. To abort the simulation which is running, the user can select Run>Abort.
The leftmost part of the screen is the tools panel. The tools panel contains sixteen objects
which can be used to build an environment, four initial robots which are the same robot but
have different rotations, and a delete button to correct a wrongly placed object. There are
sixteen objects available to build an environment. Ten of them are black lines which simulate
a black tape. The user can use these lines for example for testing a line following robot. The
other six objects are 3D obstacles. These objects can be used for testing the ultrasonic
sensor which measures the distance to the nearest object. These six obstacles are the
following: a big grey cube, a small grey cube, a big grey ball, a small grey ball and a wall
which can be placed horizontally and vertically.
The gridded panel in the centre is the surface in which the objects and the robot are placed.
Each cell in the grid represents a 10 cm by 10 cm surface. The whole grid has nine columns
and twelve rows from which it follows that this grid represents a 90 cm by 120 cm surface.
The simulation runs in this environment panel.
The black panel at the top right of the screen is the code panel. This panel shows the code of
the imported MATLAB-file. The user can click on this panel to import the MATLAB-file
instead of using the menu bar. The white panel at the bottom right is the output panel. This
panel reports the connections with the ports, the measured values of the sensors, the line of
the MATLAB code which is running, and the print statements in the MATLAB-file.
Figure 16. The simulation screen.
15
4.5. Testing
We tested our simulator with a lot of MATLAB programs and scenarios, but in this chapter we
will explain two scenarios which are used during the lab session by the students. The ports
are connected as follows: A with the left motor, B with the right motor, 1 with the left light
sensor, 2 with the right light sensor, and 3 with the ultrasonic sensor. Table 6 shows the
MATLAB codes of the two scenarios.
OpenLight(SENSOR_1, 'ACTIVE');
OpenLight(SENSOR_2, 'ACTIVE');
mA = NXTMotor('A', 'Power', 20);
mB = NXTMotor('B', 'Power', 20);
OpenUltrasonic(SENSOR_3);
m = NXTMotor('AB', 'Power', 10);
m.SendToNXT();
while(true)
val1 = GetLight(SENSOR_1);
val2 = GetLight(SENSOR_2);
minDistance = 20;
while(GetUltrasonic(SENSOR_3) > minDistance)
if (val1 > 450 && val2 > 400)
mA.Power = 20;
mB.Power = 20;
elseif (val1 < 410 && val2 > 380)
mA.Power = 0;
mB.Power = 15;
elseif (val2 < 410 && val1 > 400)
mB.Power = 0;
mA.Power = 15;
elseif (val2 < 380 && val1 < 400)
mAB.Stop('brake');
break;
end
mA.SendToNXT();
mB.SendToNXT();
end
m.Stop(‘off’);
end
Table 6. (a) MATLAB code for line follower, (b) MATLAB code for obstacle avoidance
4.5.1. Line follower
The idea of the line follower program is as follows. There are two light sensors, a left and a
right one, and two motors, also a left and a right one. Continuously, the two light sensors
measure the reflected light. This results that there are four combinations possible. If both
sensors measure white, both motors drive straight forward with the same speed (see Figure
17(a)). If the left light sensor measures black and the right one white, then the left motor
stops driving while the right motor drives with a certain speed (see Figure 17(b)). This causes
a curve to the left. The same idea holds for the case in Figure 17(c), but this causes a curve
to the other direction. If both sensors measure black, then both motor stop driving and the
program will stop running (see Figure 17(d)). The pseudo code of the line follower is as
follows.
Figure 17. (a) Left and right measure white, (b) left measures black and right measures
white, (c) left measures white and right measures black, (d) left and right measure black.
16
WHILE motor A and B are on
measure values of left and right light sensor
IF values of left and right light sensor are white THEN
drive straight forward
ELSEIF value of left light sensor is black and right is white THEN
drive to the left
ELSEIF value of left light sensor is white and right is black THEN
drive to the right
ELSEIF values of left and right light sensor are black THEN
turn off the both motors
END
The user wrote the line follower code in MATLAB shown in Table 6a. The ports of the
controller are connected as described above. After setting the connections right in the
initialization screen, the initialization is completed. The simulation screen opens, and the
connected ports are shown in Figure 18 as light blue. The user clicks on the objects and the
robot marked as blue, and adds these to the environment and positions the robot. Then, the
user adds the MATLAB code shown in Table 6a by clicking on File>Import MATLAB-file in
the menu bar (marked as yellow). After clicking on Run>Run in the menu bar which is
marked as green, the program will start running the simulation. The robot will follow the line
and will stop when it arrives at the T-junction. The user clicks on File>Quit to quit the
program.
Figure 18. The line follower simulation.
17
4.5.2. Obstacle avoidance
The idea of the obstacle avoidance program is as follows. The robot is connected to an
ultrasonic sensor. This sensor measures continuously the distance from the robot to an
obstacle. The robot drives straight forward, until it measures a value which is smaller than a
given minimum distance, and then it stops. The corresponding pseudo code of the obstacle
avoidance is as follows.
WHILE measured distance is greater than the minimum distance
drive straight forward
END
stop driving
The user has written the code obstacle avoidance in MATLAB shown in Table 6b. The
initialization is the same as in the line follower example. The simulation screen is opened and
some info about the initialization marked as light blue is shown (see Figure 19). The user
adds objects and the robot, which are marked as blue to the environment, which is marked
as red. After importing the code of Table 6b by clicking File>Import MATLAB-file, the code is
shown. This is marked as yellow. After clicking on Run>Run which is marked as green, the
robot will drive straight and stop when the ultrasonic sensor measures a distance less than
20 cm.
There are more tests done, for example with robots connected to LEDs. All the tests are
completed successfully. The simulator works as expected.
Figure 19. The obstacle avoidance simulation.
18
Chapter 5. Conclusion
This thesis described a 2D simulator for the LEGO Mindstorms NXT robotic kit controlled by
MATLAB. We named the developed simulator “NXTSim”. For developing NXTSim, the
programming language JAVA was used. The user is able to initialize the settings of the
controller, to build an environment as desired, and run a simulation controlled by a correctly
written MATLAB-code. The robot is equipped with two sensors (light and ultrasonic sensors),
and two actuators (motors and LEDs). The user is able to build the desired robot and test it in
an environment. The defined objects for building an environment are black lines, walls and
other 3D obstacles.
The developed simulator meets all the requirements. One of the requirements was that the
speed of the simulator had to be acceptable, and we are satisfied about the speed. If we
compare NXTSim to other simulators, we see that there are simulators where the user can
use more sensors (Webots, Rob. Tbx.) or even use multiple robots (Simbad). But the focus
of NXTSim is on simplicity. NXTSim will be used by students without experience in
programming. For these students, it is easier to use a simple simulator and focus on the
MATLAB code.
For more future work, we suggest to define more objects which can be used for building the
environment. Examples of these objects are objects in different colours or other geometrical
figures. Our program recognizes twenty-six different MATLAB commands. This number can
be increased which makes the simulator accept more different codes. The functions in the
menu bar are limited. This can be extended by adding more functions. The duration of this
project was two months and the team which developed this program consists of one
Computer Science student. So using another programming language and starting a whole
new project is possible.
In order to use or to extend the simulator, permission should be asked to Müttalip Küçük. The
VU University Amsterdam has the permission to use this program during the lab sessions of
the course “Pervasive Computing”. You can find more information about NXTSim on its
official page: www.mkucuk.nl/nxtsim.
Personally, my main goal of choosing this project was to improve my programming skills in
JAVA. Before this project, I had some experience which I gained during my study. But after
this project, I learned a lot about the graphical user interface components in JAVA. So my
personal goal is also completed successfully.
I thank Natalia Silvis-Cividjian for helping us with the literature study and for supervising us
during this project. I also thank Anton Eliens for being the second reader of this paper.
19
References
Aach, T., Behrens, A. (2008). Freshman Engineers Build MATLAB Powered LEGO Robots. MATLAB Digest.
Academic Edition. The MathWorks. 08/08.
Anderson, D. P., Kubiatowicz, J. (2002). The Worldwide Computer. Scientific American. March 2002.
Carnegie Mellon Robot Navigation Toolkit [Online]. Available: http://carmen.sourceforge.net/.
Corke, P. I. (1995). A computer tool for simulation and analysis: the robotics toolbox for matlab. Proceedings of
the 1995 National Conference of the Australian Robot Association, Melbourne, Australia, pp 319-330, July
1995.
Easy Rob [Online]. Available: http://www.easy-rob.com/en/easy rob.html.
Friedewald, M., Raabe, O. (2011). Ubiquitous computing: An overview of technology impacts. Telematics and
Informatics 28 (2011) 66-65.
Gazebo [Online]. Available: http://gazebosim.org
Gonçalves, J., Lima, J., Malheiros, P., Costa, P. (2009). Realistic simulation of a Lego Mindstorms NXT based
robot. 18th IEEE International Conference on Control Applications Part of 2009 IEEE Multi-conference on
Systems and Control Saint Petersburg, Russia, July 8-10, 2009.
Khan, K. S., ter Riet, G., Glanville, J., Sowden, A. J., & Kleijnen, J. (2001). Undertaking systematic review of
research on effectiveness. April, 2001.
Kitchenham, B. (2004). Procedures for Performing Systematic Reviews. Keele University Technical Report
TR/SE-0401. ISSN:1353-7776. July, 2004. Research Report. CRD Report (4 (2nd Edition)). NHS Centre for
Reviews and Dissemination, York, UK.
Kumar, A. M. (2004). Three Years of Using Robots in the Artificial Intelligence Course – Lessons Learned.
Meyer, R. M., Puehn, D. C. (2009). Simulating a LEGO Mindstorms RCX Robot in the Robotran Environment.
Proceedings of the Twenty-Second International FLAIRS Conference (2009)
Microsoft Robotic Developer Studio [Online]. Available: http://www.microsoft.com/robotics/.
MissionLab [Online]. Available: http://www.cc.portch.edu/ai/robot lab/research/MissionLab/.
Patterson-McNeill, H., Binkerd, C. L. (2001). Resources for using Lego Mindstorms. Consortium for Computing
in Small Colleges. March, 2001.
Player/Stage [Online]. Available: http://playerstage.sourceforge.net/.
Robot Operating System [Online]. Available: http://www.ros.org/wiki/.
Simbad [Online]. Available: http://simbad.sourceforge.net/.
Staranowicz, A., Mariottini, G. L. (2011). A Survey and Comparison of Commercial and Open-Source Robotic
Simulator Software. PETRA&11, May 25-27, 2011, Crete, Greece.
Tian, Y. (2007). Simulation for LEGO Mindstorms Robotics. A thesis at Lincoln University.
USARSim Matlab Toolbox [Online]. Available: http://robotics.mem.drexel.edu/USAR/index.html.
Webots [Online]. Available: http://www.cyberbotics.com/overview.
Weiser, M. (1991). The Computer for the 21st Century. Scientific American, 1991.
20
Appendix
User Manual: NXTSim
Before starting the NXTSim, you should prepare the following two things.
• At least one MATLAB file (.m) which will be the program of the simulated robot.
• The settings of the ports (1, 2, 3, 4, A, B, C) which tells the connections between
ports and sensors/actuators.
You are now ready to start NXTSim. Open NXTSim.jar. Screen 1 will appear.
Firstly, click on one of the sensor ports (1, 2, 3, or 4) which are encircled with red to connect
it with a sensor. After clicking on one of the sensor ports, screen 2 will appear which allows
you to choose one of the available sensors. After choosing a sensor, you will see a black line
between the sensor port and the chosen sensor which represents a connection.
Secondly, click on one of the actuator ports (A, B, or C) which is encircled with blue to
connect it with an actuator. After clicking on one of the actuator ports, screen 3 will appear.
After clicking on one of the available actuators, it will be connected with the selected actuator
port. Screen 4 shows an example where the controller is connected with a number of
sensors and actuators.
Screen 1
Screen 2
Screen 3
Screen 4
21
If you are completed with initialization, you can click on the OK button at the bottom right.
Screen 5 will appear. This screen consists of the following panels:
• The tools panel at the left, which is encircled as blue. This panel contains objects
which can be positioned in the environment panel.
• The environment panel at the middle, which is encircled as green. This panel
represents the environment of the simulation.
• The code panel at the top right, which is encircled as yellow. This panel shows the
MATLAB file which is imported.
• The output panel at the bottom right, which is encircled as red. This panel shows the
following information:
- the connections initialized in the previous screen.
- the values of the connected sensors, if there are any.
- the line of the MATLAB-code which is running by the simulator.
Screen 5
Screen 6
Click on the one of the objects in the tools panel, except the robot. Click on the place in the
environment panel where you want to position this selected object. This object will be
positioned. Do to the same for all the objects you want to use to build the environment.
You can delete an object or the robot by first clicking on the delete icon at he bottom left, and
then on the object or robot which you want to delete from the environment panel.
If you have placed all the objects, you have to position the robot. Click on one of the four
rotations of the robot, place it in the environment like you did with the objects in the previous
step. Now, the environment is built, and you will import the MATLAB-file.
There are two ways of doing this. The first way is to click on File>Import MATLAB-file in the
menu bar at the top of the screen. The second way is to click on the code panel. After doing
one of the two ways, you are able to browse through your files, and select the MATLAB file.
Select the MATLAB file. The code panel will show the code of the inputted MATLAB as
shown in screen 6.
Now, everything is ready and you can run the program by clicking on Run>Run in the menu
bar at the top. The simulation will start.
22
In the output panel, you can see the line of the MATLAB code which is running. If the
simulator is completed with running the MATLAB code, the output panel will show: “Running
the MATLAB-file is completed.”
If you want to stop the simulation, click on Run>Abort. The simulation will stop.
You are able to perform a new simulation with the same initialized connections. To perform a
new simulation, click on File>Clear environment. All objects and the robot will be deleted
from the environment panel. After building a new environment, you can import another
MATLAB file, like explained earlier.
To quit NXTSim, click on File>Quit in the menu bar. The program stops and will disappear.
More information on www.mkucuk.nl/nxtsim.
23
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