- Email: [email protected]
Design and Implementation of a Real-Time Autonomous Navigation System Applied to Lego Robots
Proceedings of the 3rd IFAC Conference on Proceedings of the 3rd IFAC Conference on Control Advances in Proportional-Integral-Derivative Proceedings of the 3rd IFAC Conference on Proceedings of the 3rd IFAC Conference on Control Advances in Proportional-Integral-Derivative Ghent, Belgium, May 9-11, 2018 Available online at www.sciencedirect.com Advances in Proportional-Integral-Derivative Control Advances in Proportional-Integral-Derivative Proceedings of the 3rd IFAC Conference on Control Ghent, Belgium, May 9-11, 2018 Ghent, Belgium, May 9-11, 2018 Ghent, Belgium, May 9-11, 2018 Advances in Proportional-Integral-Derivative Control Ghent, Belgium, May 9-11, 2018
ScienceDirect
IFAC PapersOnLine 51-4 (2018) 340–345
Design and Implementation of a Real-Time Design and Implementation of a Real-Time Design and Implementation of a Real-Time Design and Implementation of a Applied Real-Time Autonomous Navigation System to Autonomous Navigation System Applied to Design and Implementation of a Real-Time Autonomous Navigation System Applied Autonomous Navigation System Applied to to Lego Robots Lego Robots Autonomous Navigation System Applied to Lego Robots Lego Robots Thi Thoa Mac ∗,∗∗ CosminRobots Copot ∗∗∗ Clara M. Ionescu ∗∗ ∗,∗∗ Lego Thi Thoa Mac ∗,∗∗ Cosmin Copot ∗∗∗ ∗∗∗ Clara M. Ionescu ∗ Thi Thoa Mac ∗,∗∗ Cosmin Copot ∗∗∗ Clara M. Ionescu ∗ ∗ Thi Thoa Mac ∗,∗∗ Cosmin Copot ∗∗∗ Clara M. Ionescu ∗ Dynamical and Control Research Clara Group,M. Department Thoa Systems Mac Cosmin Copot Ionescu of ∗ Thi Systems and Control Research Group, Department of ∗ Dynamical Electrical Energy, Metals, Mechanical Constructions, and Systems, Systems and Control Research Group, Department of ∗ Dynamical Dynamical Systems and Control Research Group, Department of Electrical Energy, Metals, Mechanical Constructions, and Systems, Ghent B-9052, Belgium (e-mail: {thoa.macthi, Electrical Energy, Metals, Mechanical Constructions, and Systems, ∗Ghent University,
Dynamical Systems and B-9052, Control Research Group, Department of Electrical Energy, Metals, Mechanical Constructions, and Systems, Ghent University, Ghent Belgium (e-mail: {thoa.macthi, ClaraMihaela.Ionescu}@ugent.be). Ghent University, Ghent B-9052, Belgium (e-mail: {thoa.macthi, Electrical Energy, Metals, Mechanical Constructions, and Systems, Ghent University, Ghent B-9052, Belgium (e-mail: {thoa.macthi, ClaraMihaela.Ionescu}@ugent.be). ∗∗ of Mechanical Engineering, Hanoi University of Science and ClaraMihaela.Ionescu}@ugent.be). ∗∗ School Ghent University, Ghent B-9052, Belgium (e-mail: {thoa.macthi, ClaraMihaela.Ionescu}@ugent.be). of Mechanical Engineering, Hanoi University of Science and ∗∗ School Technology, No.1 Dai Co Viet, Vietnam School of Mechanical Engineering, Hanoi University of ∗∗ ClaraMihaela.Ionescu}@ugent.be). Engineering, Hanoi University of Science Science and and Technology, No.1 Dai Co Viet, Vietnam ∗∗∗School of Mechanical Department of Electromechanics, Op3Mech, University of Antwerp, Technology, No.1 Dai Co Viet, Vietnam ∗∗ ∗∗∗School of Mechanical Engineering, Hanoi University of Science and Technology, No.1 Dai Op3Mech, Co Viet, Vietnam of Electromechanics, University of Antwerp, ∗∗∗ Department Groenenborgerlaan 171, Antwerp, Belgium (email: of University of ∗∗∗ Department Technology, No.12020 Dai Op3Mech, Co Viet, Vietnam Department of Electromechanics, Electromechanics, Op3Mech, University of Antwerp, Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium (email: [email protected]) Groenenborgerlaan 171, Antwerp, (email: ∗∗∗ Department of Electromechanics, University of Antwerp, Groenenborgerlaan 171, 2020 2020 Op3Mech, Antwerp, Belgium Belgium (email: [email protected]) [email protected]) Groenenborgerlaan 171, 2020 Antwerp, Belgium (email: [email protected]) Abstract: Teaching [email protected]) concepts of a real-time autonomous robot system may be a Abstract: Teaching theoretical conceptssupport. of a real-time autonomous robot system mayofbe bethe challenging Teaching task without real hardware The paper discussesrobot the application Abstract: theoretical concepts of autonomous system aaa Abstract: Teaching theoretical conceptssupport. of aa real-time real-time autonomous robot system may mayofbethe challenging task without real hardware The paper discusses the application Lego Robot Teaching for teaching multi interdisciplinary subjects toautonomous Mechatronics students. real-time challenging task without real The discusses the application of Abstract: theoretical conceptssupport. of a real-time systemA a challenging task without real hardware hardware support. The paper paper discussesrobot the application ofbethe the Lego Robot for teaching multi interdisciplinary subjects to planning Mechatronics students. Amay real-time mobile robot system with perception using sensors, path algorithm, PID controller Lego Robot for teaching multi interdisciplinary subjects to Mechatronics students. A real-time challenging task without real hardware support. Thepath paper discusses the application of the Lego Robot for teaching multi interdisciplinary subjects to planning Mechatronics students. A controller real-time mobile robot system with perception using sensors, algorithm, PID is used as the case towith demonstrate the teaching methodology. Thealgorithm, novelties are introduced mobile robot system perception using sensors, path PID controller Lego Robot for teaching multi interdisciplinary subjects to planning Mechatronics students. real-time mobile robot system perception using sensors, path planning PIDAintroduced controller is used as to the case towith demonstrate the teaching methodology. Thealgorithm, novelties are compared classical robotic classes: (i) the adoption of a project-based learning approach as is used as the case to demonstrate the teaching methodology. The novelties are introduced mobile robot system with perception using sensors, path planning algorithm, PID controller is used as the case to demonstrate the teaching methodology. The novelties are introduced compared to classical classical robotic robotic classes: (i) (i) the adoption adoption of aa project-based project-based learningforapproach approach as teaching methodology; (ii) an effective real-time autonomous navigation approach the mobile compared to classes: the of learning as is used as theclassical case torobotic demonstrate teaching methodology. The novelties are introduced compared to classes:the (i) the adoption of a project-based learning approach as teaching methodology; (ii) an effective real-time autonomous navigation approach for the mobile robot. However, the extendibility and applicability of the approach are not limited to teaching methodology; (ii) real-time autonomous navigation approach for the mobile compared to classical robotic classes: (i) the adoption of presented a project-based learning as teaching methodology; (ii) an an effective effective real-time autonomous navigation approach forapproach the mobile robot. However, the extendibility extendibility and applicability applicability of the the presented approach are not not limited to only the educational purpose. robot. However, the and of presented approach are limited to teaching methodology; (ii) an effective real-time autonomous navigation approach for the mobile robot. However, the extendibility and applicability of the presented approach are not limited to only the educational purpose. only the educational purpose. robot. However, the extendibility and of theHosting presented approach to only the educational purpose. © 2018, IFAC (International Federation of applicability Automatic Control) by Elsevier Ltd. are All not rightslimited reserved. Keywords: Education, LEGO MINDSTORMS, autonomous navigation, PID control, only the educational purpose. Keywords: Education, MINDSTORMS, autonomous navigation, navigation, PID control, control, Embedded systems. LEGO Keywords: Education, Education, LEGO MINDSTORMS, MINDSTORMS, autonomous autonomous Keywords: LEGO navigation, PID PID control, Embedded systems. Embedded Education, systems. LEGO MINDSTORMS, autonomous navigation, PID control, Keywords: Embedded systems. 1. INTRODUCTION Therefore, project-based approach is a good option as it Embedded systems. 1. INTRODUCTION INTRODUCTION Therefore, approach good option as it provides to project-based students necessary skillsis fora future engineering 1. Therefore, project-based approach is good as 1. INTRODUCTION Therefore, project-based approach isforaa future good option option as it it provides to students necessary skills engineering provides to students skills engineering The design of automatic machines and robots is required career. 1. INTRODUCTION Therefore, approach good option as it provides to project-based students necessary necessary skillsisfor fora future future engineering career. The design of automatic machines and robots is required advanced frommachines mechanics, automatic control career. The designknowledge of and robots provides to students for future engineering career. In Chevalier et al., necessary (2017)), askills solution implemented at The design of automatic automatic machines and automatic robots is is required required advanced knowledge from mechanics, control and computer science. In high-level education it is there- In advanced knowledge from mechanics, automatic control Chevalier et al., (2017)), aaofsolution implemented at career. The design of automatic machines and robots is required our lab to address the problem practice for large groups In Chevalier et al., (2017)), solution implemented at advanced knowledge from mechanics, automatic control and computerto science. In activities high-leveltoeducation education it is isacquire there- In Chevalier et al.,the(2017)), aofsolution implemented at fore desirable organize let students and computer science. In high-level it thereour lab to address problem practice for large groups advanced knowledge from mechanics, automatic control students is using remote lab applications. By this apour lab to address the problem of practice for large groups and computer science. In activities high-level education it isacquire there- of fore desirable to organize to let students In Chevalier etusing al.,theremote (2017)), aofapplications. solution at ourstudents lab to address problem practice implemented for large groups more powerful competencies for the integrated design of fore desirable to organize activities to let students acquire of is lab By this apand computer science. In activities high-level itdesign isacquire therethe students are lab able to process the this experiof students is By apfore desirable to organize toeducation let students more powerful competencies for the the integrated of plications, our lab to address theremote problem ofapplications. practice for large groups of students is using using remote lab applications. By this apautomatic machines and real-time, object-oriented control more powerful competencies for integrated design of plications, the students are able to process the experifore desirable tocompetencies organize activities tointegrated let students acquire fromthe various locations. In applications. Copot et al., By (2017), we plications, students are able to the experimore powerful for the design of ments automatic machines and real-time, real-time, object-oriented control of students is using remote this applications, the students are lab able to process process the experisoftware. Recently, Lego Mindstorms, a commercial robot automatic machines and object-oriented control ments from various locations. In Copot et al., (2017), we more powerful competencies for the integrated design of introduced the academic context, and projects with the ments from various locations. In Copot et al., (2017), we automatic machines and real-time, object-oriented control software. Recently, Legotesting Mindstorms, a applications commercial robot robot plications, students are able to process experiments fromthe various locations. In Copot et al., the (2017), we kit has been used for robotics (Rosoftware. Recently, Lego Mindstorms, a commercial introduced the academic context, and projects with the automatic machines andtesting real-time, object-oriented course of ICT and Mechatronics (the full description of introduced the academic context, and projects with the software. Recently, Lego Mindstorms, a applications commercialcontrol robot kit has been used for robotics (Roments from various locations. In(the Copot et al., (2017), we introduced the academic context, and projects with the driguez et al., (2016)) and more actively implemented as kit has been used for testing robotics applications (Rocourse of ICT and Mechatronics full description of software. Recently, Legotesting Mindstorms, a applications commercial robot is ICT available online (ICT and Mechatronics (2017)) of Mechatronics (the full description of the kit has et been used for robotics (Rodriguez al.,tool (2016)) and more actively implemented as course introduced theand academic context, with of ICT and Mechatronics (theand fullprojects description of the the teaching for mechatronics and control engineering driguez et al., (2016)) and more actively implemented as course is available online (ICT and Mechatronics (2017)) kit has been used testing robotics applications (RoGhent university such(ICT as (1) design of an Anti course is available online and Mechatronics (2017)) driguez et al.,tool (2016)) and more actively implemented as at the teaching forfor mechatronics and control control engineering of and Mechatronics (the full description oflock the course is ICT available online (ICT and Mechatronics (2017)) (Kim et et al.,al.,tool (2011)). It and allows students to implemented rapidly develop the teaching for mechatronics and engineering at Ghent university such as (1) design of an Anti lock driguez (2016)) more actively as Braking System (ABS); (2) two-track-line follower system at Ghent university such as (1) design of an Anti lock the teaching tool for mechatronics and control engineering (Kim et al., (2011)). It allows students to rapidly develop course is available online (ICT and Mechatronics (2017)) at Ghent university such as (1) design of an Anti lock real-time robot applications. Lego Mindstorms have been Braking (Kim et al., al., (2011)). It students to rapidly rapidly develop (2) two-track-line system the teaching tool for mechatronics and control engineering which is System able to(ABS); avoid obstacles presentoffollower along the line; Braking System (ABS); (2) follower system (Kim et (2011)). It allows allowsLego students to develop real-time robot applications. Mindstorms have been Ghent university such astwo-track-line (1) design an Anti lock Braking System (ABS); (2) two-track-line follower system already used within the framework of programming and at real-time robot applications. Lego Mindstorms have been which is able to avoid obstacles present along the line; (Kim et al., (2011)). It allows students to rapidly develop (3) sensor fusion on a two-track system; (4) design and which is able to avoid obstacles present along the real-time robot applications. Lego Mindstorms have been already used within the framework of programming and Braking System (ABS); (2) two-track-line follower system which is able to avoid obstacles system; present along the line; line; control for robot competitions (Grandi et al., (2014)). already used within the framework of programming and (3) sensor fusion on a two-track (4) design and real-time robot applications. Lego Mindstorms have been implementing control strategy for a robot arm. (3) sensor fusion on a two-track system; (4) design and already used within the framework of programming and control for real-time robot competitions competitions (Grandi et al., al.,robot (2014)). which is able to avoid obstacles present along the line; (3) sensor fusion on astrategy two-track system; (4) design and The stable control of a segway-like with control for robot (Grandi et (2014)). implementing control for a robot arm. already used within the framework of programming and implementing control strategy for a robot arm. control for robot competitions (Grandi et al., (2014)). The stable real-time control of a segway-like robot with (3) sensor fusion on a two-track system; (4) design and implementing control strategy for a robot arm. As autonomous navigation is one of the most important rea PID controller is control used asofcase studyet toal.,robot clarify the The stable real-time segway-like with control for real-time robot competitions (Grandi (2014)). The stable ofcase aa segway-like robot with autonomous navigation is one most important rea PID controller is control used aset study topresents clarify the As implementing control strategy forof athe robot arm. quirements of an intelligent vehicle, last year, we included As autonomous navigation is one of the most important retheoretical concepts. Zdesar al., (2017) the a PID controller is used as case study to clarify The stable real-time ofcase aal.,segway-like robot As autonomous navigation isvehicle, one of the most important reatheoretical PID controller is control used aset study topresents clarify with quirements of an intelligent last year, we included concepts. Zdesar (2017) the this topic for the course of ICT and Mechatronics. The aim quirements of an intelligent vehicle, last year, we included teaching approach used at Autonomous Mobile Robots theoretical concepts. Zdesar et al., (2017) presents the As autonomous navigation is one of the most important rea PID controller is used as case study to clarify quirements of an intelligent vehicle, last year, we included theoretical concepts. Zdesar et al., (2017) presents the topic for the of ICT and Mechatronics. The aim teaching approach used at Autonomous Autonomous Mobile Robots this of this project is course to design and implement a mobile robot this topic for the course of ICT and Mechatronics. The aim course that is about used algorithms inal., autonomous navigation. teaching approach at Mobile Robots quirements of an intelligent vehicle, last year, we included theoretical concepts. Zdesar et (2017) presents the this topic for the course of ICT and Mechatronics. The aim teaching approach used at Autonomous Mobile Robots of this project is to design and implement a mobile robot course that is of about algorithms in autonomous autonomous navigation. is project able safely reach theand target while avoiding unof this is to and implement aa mobile robot In thisthat area technology, there is a gap between the that course is about algorithms in navigation. topic forto the course of ICT Mechatronics. The aim teaching approach used at Autonomous Mobile Robots of this project issafely to design design and implement mobile robot course is about algorithms in autonomous navigation. that is able to reach the target while avoiding unIn thisthat area of technology, there isaspects a gap gap between the this certain obstacles in unknown lab-scale environment. There that is able to safely reach the target while avoiding unresearchers working on theoretical and engineerIn this area of technology, there is a between the this to unknown design implement a mobile robot course about algorithms in autonomous navigation. that is project able to issafely reachand the targetenvironment. while avoiding unIn thisthat areais of technology, there isaspects a gapand between the of certain obstacles in lab-scale There researchers working on theoretical engineerare four required components of autonomous navigation: certain obstacles in unknown lab-scale environment. There ing working on problems with mechanical design, motor researchers working on theoretical aspects and engineerthat is able to safely reach the target while avoiding unIn this area of technology, there is a gap between the certain obstacles in unknown lab-scale environment. There researchers working on theoretical aspects and engineerfour required autonomous navigation: ing working on problems problems with mechanical mechanical design, motor motor i) perception, theincomponents robot uses lab-scale itsof sensors to extract meanare four required components of autonomous navigation: control, camera, sensors, computer, communication, etc. are ing working on with design, certain unknown environment. There researchers working on theoretical aspects and engineerare fourobstacles required components ofsensors autonomous navigation: ing working on problems with mechanical design, motor i) perception, the robot uses its to extract meancontrol, camera, sensors, computer, communication, etc. ingful information; ii) localization, the robot determines i) perception, the robot uses to meancontrol, camera, sensors, computer, computer, communication, etc. four required autonomous navigation: ing working on problems with mechanical design, motor i) perception, the components robot uses its itsofsensors sensors to extract extract meancontrol, camera, sensors, communication, etc. are ingful information; ii) localization, the robot determines Sponsor and financial support acknowledgment goes here. Paper its location in the working space; iii) cognition and path ingful information; ii) localization, the robot determines i) perception, the robot usesspace; its sensors to extract meancontrol, camera, sensors, computer, communication, etc. ingful information; ii) localization, thecognition robot determines Sponsor and financial support acknowledgment goes here. not Paper its location in the working iii) and path titles should be written in uppercase and lowercase letters, all Sponsor and financial support acknowledgment goes here. Paper planning, the robot decides how to steer to achieve its its location in the working space; iii) cognition and ingful information; ii)decides localization, the robot determines Sponsor and support acknowledgment goes here. not Paper its location in robot the working space; iii) cognition and path path titles should be financial written in uppercase and lowercase letters, all planning, the how to steer to achieve its uppercase. titles should be written in uppercase and lowercase letters, not all planning, the robot decides how to steer to achieve its titles should be written in uppercase and lowercase letters, not all Sponsor and financial support acknowledgment goes here. Paper its locationthe in robot the working iii)steer cognition and path uppercase. planning, decidesspace; how to to achieve its uppercase. uppercase. titles should be written in uppercase and lowercase letters, not all planning, the robot decides how to steer to achieve its
Copyright © 2018, 2018 IFAC 340Hosting by Elsevier Ltd. All rights reserved. uppercase. 2405-8963 © IFAC (International Federation of Automatic Control) Copyright © 2018 IFAC 340 Copyright 2018 responsibility IFAC 340Control. Peer review©under of International Federation of Automatic Copyright © 2018 IFAC 340 10.1016/j.ifacol.2018.06.088 Copyright © 2018 IFAC 340
IFAC PID 2018 Ghent, Belgium, May 9-11, 2018
Thi Thoa Mac et al. / IFAC PapersOnLine 51-4 (2018) 340–345
341
close to the ground, in this way even very low obstacles can be detected. At the starting point the robot turns a certain angle and meanwhile measures the distances to the nearest object using the ultrasonic sensor. In this way the robot can detect all the obstacles around it. The data from the ultrasonic sensor and the angle at which this distance is measured can be transformed to (x, y) coordinates. The algorithm Fig. 1. The Lego robot design, (a) side view of the robot; (b) the front view of the robot. goal; iv) motion control, the robot regulates its motion to accomplish the desired trajectory (Mac et al., (2016); Mac, (2016)). Therefore, in this project, students have to design mobile robot, choose appropriate sensor perception, develop suitable path planning and controller. The following innovative aspects are introduced with respect to classical robotic classes: (i) the use of Lego robots as teaching tools; (ii) the adoption of a project-based learning method as teaching methodology; (iii) providing an effective real-time autonomous navigation approach for mobile robot. The remainder of this paper is structured as follows. In section 2, a brief introduction of mobile robot platform and obstacle detection method are presented. Section 3 describes the path planning algorithm while section 4 presents PID controller. The experiment is demonstrated in section 5, follow by section 6, where the main outcomes of this work are summarized and the future works are discussed. 2. THE MOBILE ROBOT PLATFORM AND OBSTACLE DETECTION METHOD In this section, we introduce about robot design system and obstacles detection of a mobile platform. 2.1 Robot design The robot design is shown in Fig 1 with two driven wheels. The third supporting wheel is a non-driven castor that follows the movement imposed by the two driven wheels. The wheel diameter is 5.5 cm. Both wheels are driven by EV3 Large Servo Motors with tacho feedback. In front of the robot, an ultrasonic sensor is mounted. The EV3 ultrasonic sensor can measure distances by emitting and detecting sound waves. It has a measuring range from 1 to 250 cm with an accuracy of ±1 cm. On the left side of the brick, an EV3 Medium Servo Motor is mounted, with on top an EV3 Infrared Sensor. This composition makes it possible to rotate the infrared sensor 3600 . This sensor is able to detect an infrared signal emitted by a remote control infrared beacon. The EV3 intelligent brick is the heart and brain of the robot. It supports USB, Bluetooth and Wi-Fi communication with a computer.
Fig. 2. Algorithm of detecting obstacles for detecting obstacles is depicted in Fig 2, where θ is the total angle the robot turns while scanning and r is the distance measured by the sensor. The obstacle lines are then made into rectangles using the LineT oRectangle class. The way of detecting obstacles and making the obstacle lines is shown in Fig 3 when robot turns for θ degrees, while measuring every ∆θ. The ultrasonic sensor (red) in front of the robot performs measurements (blue dashed lines) and stores the calculated obstacle points (yellow dots) as a (x, y) coordinates. Obstacle points that are separated a distance ∆r in radius smaller than a given value, are connected to obstacle lines (yellow lines connecting obstacle points). Those yellow lines will later be converted to rectangles as shown in Fig 4. At the left, the obstacles (green) with real dimensions are shown. Taking into account the robot dimension, the obstacles are extended as shown in the right. 3. PATH PLANNING ALGORITHM For this project, the students are advised to choose one basic path planning method such as A*, D*, or shortest
2.2 Detecting obstacles using ultrasonic sensor The robot has to plan a path to a destination point, so it needs to know where the obstacles are. For detecting the obstacles, an ultrasonic sensor is chosen because this sensor measures the distances accurately. The sensor is placed 341
Fig. 3. Object detection and interpretation
IFAC PID 2018 342 Ghent, Belgium, May 9-11, 2018
Thi Thoa Mac et al. / IFAC PapersOnLine 51-4 (2018) 340–345
Fig. 4. Visualisation of LineMap formation
Fig. 6. The cost function graph of the shortest path finder method where G is the cost from the start node the present node. H is the heuristic that estimate the cost of the cheapest path from that node to the goal. The main difference is that shortest path finder does not use a grid but uses its start and goal location, and the end points of all obstacles to construct a graph in the map. The algorithm is presented in Fig 5. To show the concept, one example is depicted in Fig 6. From the start location, the possible paths are from start to node 1 or from start to node 2. The F -value from start to node 1 is calculated as the sum of its G-value (see Table 1), and the H-value of node 1 (as shown in Fig 6). Table 1. G-values for all possible edges Path start → 1 start → 2 1→2 2→3 2→4 2→5 2→6 3→4 3→5 3→6 4→5 6 → target
G - value 4,410 7,210 7,000 4,000 13,42 12,65 13,60 10,82 9,850 10,77 1,000 3,160
Table 2. The F cost value
Fig. 5. Shortest path finder algorithm flowchart. path finder method. In this section, we introduce about shortest path finder method which used in the leJOS programming language. Similar to the A* algorithm, the path is found by searching among all possible paths to the target the one that has smallest cost. The cost function of the shortest finder method is given in (1) F =G+H (1) 342
Path start start start start start start start start
→1 →2 →1→2 →2→3 →2→4 →2→5 →2→5 → 2 → 6 → target
F -value 23,82 23,97 28,38 25,13 24,75 23,98 23,97 23,97
As the F -value for the path between start and node 1, is lower than the one between start and node 2, node 1 is chosen for being added to the closed list. Next only one path is possible starting from node 1, namely the path to node 2. But similarly as explained for the A* algorithm,
IFAC PID 2018 Ghent, Belgium, May 9-11, 2018
qr
e −
Thi Thoa Mac et al. / IFAC PapersOnLine 51-4 (2018) 340–345
[vR , vL ]T PID Controller
Lego robot
343
q
qm Sensor feedbacks Fig. 7. Mobile robot control system.
Fig. 8. The Kinematic model of the Lego robot. the F -value when traveling directly from start to node 2 is lower than the one when traveling along the combined path start-node 1-node 2, as indicated in Table 2. So consequently node 1 is deleted of closed and is replaced by node 2. Next, four paths are possible for the robot to travel with node 2 as a starting point namely from node 2 to node 3, node 4, node 5 and to node 6. The F -values for these paths are again given in Table 2. As can be seen in the table, the next node that will added to the closed list will be node 6, as the F -value from node 2 to node 6 is the lowest of the four paths starting from node 2. Finally the only path possible starting from node 6 is the path to the target location, and with that making an end to the shortest path search. 4. MOBILE ROBOT CONTROL SYSTEM
Fig. 9. Scanning unknown obstacles with ultrasonic sensor x˙ r cos θ r cos θ y˙ = r sin θ r sin θ vR (3) vL r r −D θ˙ D 5. EXPERIMENTAL RESULTS
5.1 Sensor accuracy To decide which sensor to use to measure distances, an experiment has been performed. Both the ultrasonic sensor and the infrared sensor were tested. A comparison between real and measured values is shown in Table 3. It is clear that the ultrasonic sensor is the most accurate one: ± 1cm. The measured distances of the infrared sensor seems to be different depending on the material of the object measured. Therefore, the ultrasonic sensor is used for obstacle detection in this project. Table 3. The Distance Measurement
To this day, over 95 percentage of industrial applications are predominantly controlled by (PID) controllers due to is simplification and accuracy. A PID controller includes: Ki GP ID (s) = Kp + + Kd s (2) s where Kp , Ki , Kd denote the coefficients for the proportional, integral, and derivative terms. The proposed control system for the Lego robot in this study is shown in Fig 7. The control system consists of a PID controller, the kinematic model of mobile robot, a reference trajectory generator and two encoders which provides odometric information. In which vL and vR are the speed of the left and right wheel in cm/s respectively. Input, output, measured parameters are vectors: qr =[xr , yr , θr ]T , q =[x, y, θ]T , qm =[xm , ym , θm ]T . A schematic of the kinematic model is shown in Fig 8. The kinematic model is presented in equation (3). The parameters D and r are respectively the distance between two wheels and the wheel radius. 343
Real Distance 0 15 30 45 60 75 90 105 120
Infrared sensor black 0 18 34 44 51 ∞ ∞ ∞ ∞
Infrared sensor color 0 24 40 48 ∞ ∞ ∞ ∞ ∞
Ultrasonic Sensor 0 16.9 30.2 45.6 60.2 75.1 91.1 105.0 120.1
The algorithm used to detect unknown obstacles, is described in Section 2. In the lab-scale environment, three known obstacles were placed around the robot and their actual positions were compared to the positions measured by the robot’s sensor. Fig 10 shows the results of this test. These results are satisfactory, as the measured position is very close to the real obstacle.
IFAC PID 2018 344 Ghent, Belgium, May 9-11, 2018
Thi Thoa Mac et al. / IFAC PapersOnLine 51-4 (2018) 340–345
Fig. 11. Landing in presence of known/unknown obstacles results in the sequence of time 5.3 Autonomous navigation experiment In this experiments, Lego Mindstorms EV3 was moving in presence of uncertain obstacles. The yellow long wall is a known obstacle, while the red and blue walls are unknown obstacles. The start point and target point are marked on the map. Experiments were carried out to test the ability of the robot performing autonomous task. As soon as the robot detected unknown obstacle by ultrasonic sensor, the position of that obstacle is provided. The results show that the Lego robot avoid all obstacles on its path and stop on the target position. The intended accuracy has been achieved using an equipped ultrasonic sensor. 6. CONCLUSION
Fig. 10. (a.) Optimal path calculated by Shortest Path Finder method taking into account the robot size red path; (b.) compare with the path obtained by A* algorithm - pink path; (c.) Real path experiment by Shortest Path Finder method. 5.2 Shortest path finder experiment In this experiment, we evaluated the shortest path finder method by simulation and experiment. We also compared it with A* algorithm. Fig 10 presents the path obtained by shortest path finder algorithm - red path (a.); this path is compared to the one obtained by A* algorithm - pink path (b.) and the real path obtained by experiment - blue path (c.). The shortest path finder algorithm’s path is shorter than A* algorithm’s path. Moreover, the ShortestP athF inder only uses the endpoints of obstacles as nodes, it has a much shorter calculation time, as the amount of endpoints of obstacles is usually much smaller than the amount of cells in a grid used in A* algorithm. It is also able to pass closer to obstacles than A*, as A* uses the middle of a grid cell as a way-point. This is a reason the ShortestP athF inder algorithm is suggested in this project. Fig. 10 c shows that the robot is able to track the designed path accurately. 344
The design of automatics machine and robot is a discipline that needs practical work at the learning process. To carry out it at the laboratory with the students, the LEGO system is proposed. This platform provides basically 2 advantages: its low price and good features. This study proposed an effective educational hands-on testbed so that students can learn a concept and design process for using sensor system, path planning, and control processing. This aim of the added project is making an autonomous robot that is able to drive to a goal while avoiding known and unknown obstacles. In paper, we introduced ShortestP athF inder algorithm, which is a modified version of the A* algorithm. This comparison results showed the ShortestP athF inder algorithm to be the superior to the A* algorithm. The ultrasonic sensor was implemented to detect unknown obstacles, as it measures distances very accurately. Detecting unknown obstacles and determining their size was also done by using the ultrasonic sensor. The Lego robot was able to autonomously navigate in an unknown environment. The results of the experiments were satisfying. The achievements of our teaching method compared to classical robotic classes are: (i) the applications of a project-based learning approach as teaching methodology; (ii) the development of an effective real-time autonomous navigation approach for mobile robot. REFERENCES Rodriguez, C., Guzman, J. L., Berenguel, M., Dormido, S. (2016), Teaching real-time programming using mobile
IFAC PID 2018 Ghent, Belgium, May 9-11, 2018
Thi Thoa Mac et al. / IFAC PapersOnLine 51-4 (2018) 340–345
robots, IFAC-PapersOnLine vol. 49 (6), pp. 10-15. Kim, S., Oh, H., Kolaman, A., Tsourdos, A., White, B., Guterman, H. (2011), Educational hands-on test bed using Lego robot for learning guidance, navigation, and control, IFAC Proceedings Volumes, vol. 44(1), pp. 5170-5175. Grandi, R., Falconi, R., Melchiorri, (2014), Robotic Competitions: Teaching Robotics and Real-Time Programming with LEGO Mindstorms, IFAC Proceedings Volumes, vol. 47(3), pp. 10598-10603. Zdesar, A., Blazic, S., Klancar, G. (2017), Engineering Education in Wheeled Mobile Robotics, IFAC Proceedings Volumes, vol. 50(1), pp. 12173-12178. Chevalier, A., Copot, C., Ionescu, C., DeKeyser, R. (2017), A three-year feedback study of a remote laboratory used in control engineering studies, IEEE Transactions on Education, vol. 60(2), pp.127-133.
345
345
Copot, C., Ionescu, C., DeKeyser, R. (2017), Interdisciplinary project-based learning at master level: control of robotic mechatronic IFAC-PapersOnLine, vol. 49(6), pp. 314319. ICT and Mechatronics course content, Ghent University, https://studiegids.ugent.be/2016/EN/studiefiches/ E019330.pdf Mac, T. T., Copot, C., Tran, D .T., De Keyser, R. (2016), Heuristic approaches in robot path planning: a surveys, Autonomous and Robotics System, vol. 86, pp. 13-28. Mac, T. T., Copot, C., Hernandez, A., De Keyser, R. (2016), Improved potential field method for unknown obstacle avoidance using UAV in indoor environment, IEEE Intentional Symposium on Applied Machine Intelligence and Informatics, Herlany, Slovakia, pp. 345350.
ScienceDirect
IFAC PapersOnLine 51-4 (2018) 340–345
Design and Implementation of a Real-Time Design and Implementation of a Real-Time Design and Implementation of a Real-Time Design and Implementation of a Applied Real-Time Autonomous Navigation System to Autonomous Navigation System Applied to Design and Implementation of a Real-Time Autonomous Navigation System Applied Autonomous Navigation System Applied to to Lego Robots Lego Robots Autonomous Navigation System Applied to Lego Robots Lego Robots Thi Thoa Mac ∗,∗∗ CosminRobots Copot ∗∗∗ Clara M. Ionescu ∗∗ ∗,∗∗ Lego Thi Thoa Mac ∗,∗∗ Cosmin Copot ∗∗∗ ∗∗∗ Clara M. Ionescu ∗ Thi Thoa Mac ∗,∗∗ Cosmin Copot ∗∗∗ Clara M. Ionescu ∗ ∗ Thi Thoa Mac ∗,∗∗ Cosmin Copot ∗∗∗ Clara M. Ionescu ∗ Dynamical and Control Research Clara Group,M. Department Thoa Systems Mac Cosmin Copot Ionescu of ∗ Thi Systems and Control Research Group, Department of ∗ Dynamical Electrical Energy, Metals, Mechanical Constructions, and Systems, Systems and Control Research Group, Department of ∗ Dynamical Dynamical Systems and Control Research Group, Department of Electrical Energy, Metals, Mechanical Constructions, and Systems, Ghent B-9052, Belgium (e-mail: {thoa.macthi, Electrical Energy, Metals, Mechanical Constructions, and Systems, ∗Ghent University,
Dynamical Systems and B-9052, Control Research Group, Department of Electrical Energy, Metals, Mechanical Constructions, and Systems, Ghent University, Ghent Belgium (e-mail: {thoa.macthi, ClaraMihaela.Ionescu}@ugent.be). Ghent University, Ghent B-9052, Belgium (e-mail: {thoa.macthi, Electrical Energy, Metals, Mechanical Constructions, and Systems, Ghent University, Ghent B-9052, Belgium (e-mail: {thoa.macthi, ClaraMihaela.Ionescu}@ugent.be). ∗∗ of Mechanical Engineering, Hanoi University of Science and ClaraMihaela.Ionescu}@ugent.be). ∗∗ School Ghent University, Ghent B-9052, Belgium (e-mail: {thoa.macthi, ClaraMihaela.Ionescu}@ugent.be). of Mechanical Engineering, Hanoi University of Science and ∗∗ School Technology, No.1 Dai Co Viet, Vietnam School of Mechanical Engineering, Hanoi University of ∗∗ ClaraMihaela.Ionescu}@ugent.be). Engineering, Hanoi University of Science Science and and Technology, No.1 Dai Co Viet, Vietnam ∗∗∗School of Mechanical Department of Electromechanics, Op3Mech, University of Antwerp, Technology, No.1 Dai Co Viet, Vietnam ∗∗ ∗∗∗School of Mechanical Engineering, Hanoi University of Science and Technology, No.1 Dai Op3Mech, Co Viet, Vietnam of Electromechanics, University of Antwerp, ∗∗∗ Department Groenenborgerlaan 171, Antwerp, Belgium (email: of University of ∗∗∗ Department Technology, No.12020 Dai Op3Mech, Co Viet, Vietnam Department of Electromechanics, Electromechanics, Op3Mech, University of Antwerp, Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium (email: [email protected]) Groenenborgerlaan 171, Antwerp, (email: ∗∗∗ Department of Electromechanics, University of Antwerp, Groenenborgerlaan 171, 2020 2020 Op3Mech, Antwerp, Belgium Belgium (email: [email protected]) [email protected]) Groenenborgerlaan 171, 2020 Antwerp, Belgium (email: [email protected]) Abstract: Teaching [email protected]) concepts of a real-time autonomous robot system may be a Abstract: Teaching theoretical conceptssupport. of a real-time autonomous robot system mayofbe bethe challenging Teaching task without real hardware The paper discussesrobot the application Abstract: theoretical concepts of autonomous system aaa Abstract: Teaching theoretical conceptssupport. of aa real-time real-time autonomous robot system may mayofbethe challenging task without real hardware The paper discusses the application Lego Robot Teaching for teaching multi interdisciplinary subjects toautonomous Mechatronics students. real-time challenging task without real The discusses the application of Abstract: theoretical conceptssupport. of a real-time systemA a challenging task without real hardware hardware support. The paper paper discussesrobot the application ofbethe the Lego Robot for teaching multi interdisciplinary subjects to planning Mechatronics students. Amay real-time mobile robot system with perception using sensors, path algorithm, PID controller Lego Robot for teaching multi interdisciplinary subjects to Mechatronics students. A real-time challenging task without real hardware support. Thepath paper discusses the application of the Lego Robot for teaching multi interdisciplinary subjects to planning Mechatronics students. A controller real-time mobile robot system with perception using sensors, algorithm, PID is used as the case towith demonstrate the teaching methodology. Thealgorithm, novelties are introduced mobile robot system perception using sensors, path PID controller Lego Robot for teaching multi interdisciplinary subjects to planning Mechatronics students. real-time mobile robot system perception using sensors, path planning PIDAintroduced controller is used as to the case towith demonstrate the teaching methodology. Thealgorithm, novelties are compared classical robotic classes: (i) the adoption of a project-based learning approach as is used as the case to demonstrate the teaching methodology. The novelties are introduced mobile robot system with perception using sensors, path planning algorithm, PID controller is used as the case to demonstrate the teaching methodology. The novelties are introduced compared to classical classical robotic robotic classes: (i) (i) the adoption adoption of aa project-based project-based learningforapproach approach as teaching methodology; (ii) an effective real-time autonomous navigation approach the mobile compared to classes: the of learning as is used as theclassical case torobotic demonstrate teaching methodology. The novelties are introduced compared to classes:the (i) the adoption of a project-based learning approach as teaching methodology; (ii) an effective real-time autonomous navigation approach for the mobile robot. However, the extendibility and applicability of the approach are not limited to teaching methodology; (ii) real-time autonomous navigation approach for the mobile compared to classical robotic classes: (i) the adoption of presented a project-based learning as teaching methodology; (ii) an an effective effective real-time autonomous navigation approach forapproach the mobile robot. However, the extendibility extendibility and applicability applicability of the the presented approach are not not limited to only the educational purpose. robot. However, the and of presented approach are limited to teaching methodology; (ii) an effective real-time autonomous navigation approach for the mobile robot. However, the extendibility and applicability of the presented approach are not limited to only the educational purpose. only the educational purpose. robot. However, the extendibility and of theHosting presented approach to only the educational purpose. © 2018, IFAC (International Federation of applicability Automatic Control) by Elsevier Ltd. are All not rightslimited reserved. Keywords: Education, LEGO MINDSTORMS, autonomous navigation, PID control, only the educational purpose. Keywords: Education, MINDSTORMS, autonomous navigation, navigation, PID control, control, Embedded systems. LEGO Keywords: Education, Education, LEGO MINDSTORMS, MINDSTORMS, autonomous autonomous Keywords: LEGO navigation, PID PID control, Embedded systems. Embedded Education, systems. LEGO MINDSTORMS, autonomous navigation, PID control, Keywords: Embedded systems. 1. INTRODUCTION Therefore, project-based approach is a good option as it Embedded systems. 1. INTRODUCTION INTRODUCTION Therefore, approach good option as it provides to project-based students necessary skillsis fora future engineering 1. Therefore, project-based approach is good as 1. INTRODUCTION Therefore, project-based approach isforaa future good option option as it it provides to students necessary skills engineering provides to students skills engineering The design of automatic machines and robots is required career. 1. INTRODUCTION Therefore, approach good option as it provides to project-based students necessary necessary skillsisfor fora future future engineering career. The design of automatic machines and robots is required advanced frommachines mechanics, automatic control career. The designknowledge of and robots provides to students for future engineering career. In Chevalier et al., necessary (2017)), askills solution implemented at The design of automatic automatic machines and automatic robots is is required required advanced knowledge from mechanics, control and computer science. In high-level education it is there- In advanced knowledge from mechanics, automatic control Chevalier et al., (2017)), aaofsolution implemented at career. The design of automatic machines and robots is required our lab to address the problem practice for large groups In Chevalier et al., (2017)), solution implemented at advanced knowledge from mechanics, automatic control and computerto science. In activities high-leveltoeducation education it is isacquire there- In Chevalier et al.,the(2017)), aofsolution implemented at fore desirable organize let students and computer science. In high-level it thereour lab to address problem practice for large groups advanced knowledge from mechanics, automatic control students is using remote lab applications. By this apour lab to address the problem of practice for large groups and computer science. In activities high-level education it isacquire there- of fore desirable to organize to let students In Chevalier etusing al.,theremote (2017)), aofapplications. solution at ourstudents lab to address problem practice implemented for large groups more powerful competencies for the integrated design of fore desirable to organize activities to let students acquire of is lab By this apand computer science. In activities high-level itdesign isacquire therethe students are lab able to process the this experiof students is By apfore desirable to organize toeducation let students more powerful competencies for the the integrated of plications, our lab to address theremote problem ofapplications. practice for large groups of students is using using remote lab applications. By this apautomatic machines and real-time, object-oriented control more powerful competencies for integrated design of plications, the students are able to process the experifore desirable tocompetencies organize activities tointegrated let students acquire fromthe various locations. In applications. Copot et al., By (2017), we plications, students are able to the experimore powerful for the design of ments automatic machines and real-time, real-time, object-oriented control of students is using remote this applications, the students are lab able to process process the experisoftware. Recently, Lego Mindstorms, a commercial robot automatic machines and object-oriented control ments from various locations. In Copot et al., (2017), we more powerful competencies for the integrated design of introduced the academic context, and projects with the ments from various locations. In Copot et al., (2017), we automatic machines and real-time, object-oriented control software. Recently, Legotesting Mindstorms, a applications commercial robot robot plications, students are able to process experiments fromthe various locations. In Copot et al., the (2017), we kit has been used for robotics (Rosoftware. Recently, Lego Mindstorms, a commercial introduced the academic context, and projects with the automatic machines andtesting real-time, object-oriented course of ICT and Mechatronics (the full description of introduced the academic context, and projects with the software. Recently, Lego Mindstorms, a applications commercialcontrol robot kit has been used for robotics (Roments from various locations. In(the Copot et al., (2017), we introduced the academic context, and projects with the driguez et al., (2016)) and more actively implemented as kit has been used for testing robotics applications (Rocourse of ICT and Mechatronics full description of software. Recently, Legotesting Mindstorms, a applications commercial robot is ICT available online (ICT and Mechatronics (2017)) of Mechatronics (the full description of the kit has et been used for robotics (Rodriguez al.,tool (2016)) and more actively implemented as course introduced theand academic context, with of ICT and Mechatronics (theand fullprojects description of the the teaching for mechatronics and control engineering driguez et al., (2016)) and more actively implemented as course is available online (ICT and Mechatronics (2017)) kit has been used testing robotics applications (RoGhent university such(ICT as (1) design of an Anti course is available online and Mechatronics (2017)) driguez et al.,tool (2016)) and more actively implemented as at the teaching forfor mechatronics and control control engineering of and Mechatronics (the full description oflock the course is ICT available online (ICT and Mechatronics (2017)) (Kim et et al.,al.,tool (2011)). It and allows students to implemented rapidly develop the teaching for mechatronics and engineering at Ghent university such as (1) design of an Anti lock driguez (2016)) more actively as Braking System (ABS); (2) two-track-line follower system at Ghent university such as (1) design of an Anti lock the teaching tool for mechatronics and control engineering (Kim et al., (2011)). It allows students to rapidly develop course is available online (ICT and Mechatronics (2017)) at Ghent university such as (1) design of an Anti lock real-time robot applications. Lego Mindstorms have been Braking (Kim et al., al., (2011)). It students to rapidly rapidly develop (2) two-track-line system the teaching tool for mechatronics and control engineering which is System able to(ABS); avoid obstacles presentoffollower along the line; Braking System (ABS); (2) follower system (Kim et (2011)). It allows allowsLego students to develop real-time robot applications. Mindstorms have been Ghent university such astwo-track-line (1) design an Anti lock Braking System (ABS); (2) two-track-line follower system already used within the framework of programming and at real-time robot applications. Lego Mindstorms have been which is able to avoid obstacles present along the line; (Kim et al., (2011)). It allows students to rapidly develop (3) sensor fusion on a two-track system; (4) design and which is able to avoid obstacles present along the real-time robot applications. Lego Mindstorms have been already used within the framework of programming and Braking System (ABS); (2) two-track-line follower system which is able to avoid obstacles system; present along the line; line; control for robot competitions (Grandi et al., (2014)). already used within the framework of programming and (3) sensor fusion on a two-track (4) design and real-time robot applications. Lego Mindstorms have been implementing control strategy for a robot arm. (3) sensor fusion on a two-track system; (4) design and already used within the framework of programming and control for real-time robot competitions competitions (Grandi et al., al.,robot (2014)). which is able to avoid obstacles present along the line; (3) sensor fusion on astrategy two-track system; (4) design and The stable control of a segway-like with control for robot (Grandi et (2014)). implementing control for a robot arm. already used within the framework of programming and implementing control strategy for a robot arm. control for robot competitions (Grandi et al., (2014)). The stable real-time control of a segway-like robot with (3) sensor fusion on a two-track system; (4) design and implementing control strategy for a robot arm. As autonomous navigation is one of the most important rea PID controller is control used asofcase studyet toal.,robot clarify the The stable real-time segway-like with control for real-time robot competitions (Grandi (2014)). The stable ofcase aa segway-like robot with autonomous navigation is one most important rea PID controller is control used aset study topresents clarify the As implementing control strategy forof athe robot arm. quirements of an intelligent vehicle, last year, we included As autonomous navigation is one of the most important retheoretical concepts. Zdesar al., (2017) the a PID controller is used as case study to clarify The stable real-time ofcase aal.,segway-like robot As autonomous navigation isvehicle, one of the most important reatheoretical PID controller is control used aset study topresents clarify with quirements of an intelligent last year, we included concepts. Zdesar (2017) the this topic for the course of ICT and Mechatronics. The aim quirements of an intelligent vehicle, last year, we included teaching approach used at Autonomous Mobile Robots theoretical concepts. Zdesar et al., (2017) presents the As autonomous navigation is one of the most important rea PID controller is used as case study to clarify quirements of an intelligent vehicle, last year, we included theoretical concepts. Zdesar et al., (2017) presents the topic for the of ICT and Mechatronics. The aim teaching approach used at Autonomous Autonomous Mobile Robots this of this project is course to design and implement a mobile robot this topic for the course of ICT and Mechatronics. The aim course that is about used algorithms inal., autonomous navigation. teaching approach at Mobile Robots quirements of an intelligent vehicle, last year, we included theoretical concepts. Zdesar et (2017) presents the this topic for the course of ICT and Mechatronics. The aim teaching approach used at Autonomous Mobile Robots of this project is to design and implement a mobile robot course that is of about algorithms in autonomous autonomous navigation. is project able safely reach theand target while avoiding unof this is to and implement aa mobile robot In thisthat area technology, there is a gap between the that course is about algorithms in navigation. topic forto the course of ICT Mechatronics. The aim teaching approach used at Autonomous Mobile Robots of this project issafely to design design and implement mobile robot course is about algorithms in autonomous navigation. that is able to reach the target while avoiding unIn thisthat area of technology, there isaspects a gap gap between the this certain obstacles in unknown lab-scale environment. There that is able to safely reach the target while avoiding unresearchers working on theoretical and engineerIn this area of technology, there is a between the this to unknown design implement a mobile robot course about algorithms in autonomous navigation. that is project able to issafely reachand the targetenvironment. while avoiding unIn thisthat areais of technology, there isaspects a gapand between the of certain obstacles in lab-scale There researchers working on theoretical engineerare four required components of autonomous navigation: certain obstacles in unknown lab-scale environment. There ing working on problems with mechanical design, motor researchers working on theoretical aspects and engineerthat is able to safely reach the target while avoiding unIn this area of technology, there is a gap between the certain obstacles in unknown lab-scale environment. There researchers working on theoretical aspects and engineerfour required autonomous navigation: ing working on problems problems with mechanical mechanical design, motor motor i) perception, theincomponents robot uses lab-scale itsof sensors to extract meanare four required components of autonomous navigation: control, camera, sensors, computer, communication, etc. are ing working on with design, certain unknown environment. There researchers working on theoretical aspects and engineerare fourobstacles required components ofsensors autonomous navigation: ing working on problems with mechanical design, motor i) perception, the robot uses its to extract meancontrol, camera, sensors, computer, communication, etc. ingful information; ii) localization, the robot determines i) perception, the robot uses to meancontrol, camera, sensors, computer, computer, communication, etc. four required autonomous navigation: ing working on problems with mechanical design, motor i) perception, the components robot uses its itsofsensors sensors to extract extract meancontrol, camera, sensors, communication, etc. are ingful information; ii) localization, the robot determines Sponsor and financial support acknowledgment goes here. Paper its location in the working space; iii) cognition and path ingful information; ii) localization, the robot determines i) perception, the robot usesspace; its sensors to extract meancontrol, camera, sensors, computer, communication, etc. ingful information; ii) localization, thecognition robot determines Sponsor and financial support acknowledgment goes here. not Paper its location in the working iii) and path titles should be written in uppercase and lowercase letters, all Sponsor and financial support acknowledgment goes here. Paper planning, the robot decides how to steer to achieve its its location in the working space; iii) cognition and ingful information; ii)decides localization, the robot determines Sponsor and support acknowledgment goes here. not Paper its location in robot the working space; iii) cognition and path path titles should be financial written in uppercase and lowercase letters, all planning, the how to steer to achieve its uppercase. titles should be written in uppercase and lowercase letters, not all planning, the robot decides how to steer to achieve its titles should be written in uppercase and lowercase letters, not all Sponsor and financial support acknowledgment goes here. Paper its locationthe in robot the working iii)steer cognition and path uppercase. planning, decidesspace; how to to achieve its uppercase. uppercase. titles should be written in uppercase and lowercase letters, not all planning, the robot decides how to steer to achieve its
Copyright © 2018, 2018 IFAC 340Hosting by Elsevier Ltd. All rights reserved. uppercase. 2405-8963 © IFAC (International Federation of Automatic Control) Copyright © 2018 IFAC 340 Copyright 2018 responsibility IFAC 340Control. Peer review©under of International Federation of Automatic Copyright © 2018 IFAC 340 10.1016/j.ifacol.2018.06.088 Copyright © 2018 IFAC 340
IFAC PID 2018 Ghent, Belgium, May 9-11, 2018
Thi Thoa Mac et al. / IFAC PapersOnLine 51-4 (2018) 340–345
341
close to the ground, in this way even very low obstacles can be detected. At the starting point the robot turns a certain angle and meanwhile measures the distances to the nearest object using the ultrasonic sensor. In this way the robot can detect all the obstacles around it. The data from the ultrasonic sensor and the angle at which this distance is measured can be transformed to (x, y) coordinates. The algorithm Fig. 1. The Lego robot design, (a) side view of the robot; (b) the front view of the robot. goal; iv) motion control, the robot regulates its motion to accomplish the desired trajectory (Mac et al., (2016); Mac, (2016)). Therefore, in this project, students have to design mobile robot, choose appropriate sensor perception, develop suitable path planning and controller. The following innovative aspects are introduced with respect to classical robotic classes: (i) the use of Lego robots as teaching tools; (ii) the adoption of a project-based learning method as teaching methodology; (iii) providing an effective real-time autonomous navigation approach for mobile robot. The remainder of this paper is structured as follows. In section 2, a brief introduction of mobile robot platform and obstacle detection method are presented. Section 3 describes the path planning algorithm while section 4 presents PID controller. The experiment is demonstrated in section 5, follow by section 6, where the main outcomes of this work are summarized and the future works are discussed. 2. THE MOBILE ROBOT PLATFORM AND OBSTACLE DETECTION METHOD In this section, we introduce about robot design system and obstacles detection of a mobile platform. 2.1 Robot design The robot design is shown in Fig 1 with two driven wheels. The third supporting wheel is a non-driven castor that follows the movement imposed by the two driven wheels. The wheel diameter is 5.5 cm. Both wheels are driven by EV3 Large Servo Motors with tacho feedback. In front of the robot, an ultrasonic sensor is mounted. The EV3 ultrasonic sensor can measure distances by emitting and detecting sound waves. It has a measuring range from 1 to 250 cm with an accuracy of ±1 cm. On the left side of the brick, an EV3 Medium Servo Motor is mounted, with on top an EV3 Infrared Sensor. This composition makes it possible to rotate the infrared sensor 3600 . This sensor is able to detect an infrared signal emitted by a remote control infrared beacon. The EV3 intelligent brick is the heart and brain of the robot. It supports USB, Bluetooth and Wi-Fi communication with a computer.
Fig. 2. Algorithm of detecting obstacles for detecting obstacles is depicted in Fig 2, where θ is the total angle the robot turns while scanning and r is the distance measured by the sensor. The obstacle lines are then made into rectangles using the LineT oRectangle class. The way of detecting obstacles and making the obstacle lines is shown in Fig 3 when robot turns for θ degrees, while measuring every ∆θ. The ultrasonic sensor (red) in front of the robot performs measurements (blue dashed lines) and stores the calculated obstacle points (yellow dots) as a (x, y) coordinates. Obstacle points that are separated a distance ∆r in radius smaller than a given value, are connected to obstacle lines (yellow lines connecting obstacle points). Those yellow lines will later be converted to rectangles as shown in Fig 4. At the left, the obstacles (green) with real dimensions are shown. Taking into account the robot dimension, the obstacles are extended as shown in the right. 3. PATH PLANNING ALGORITHM For this project, the students are advised to choose one basic path planning method such as A*, D*, or shortest
2.2 Detecting obstacles using ultrasonic sensor The robot has to plan a path to a destination point, so it needs to know where the obstacles are. For detecting the obstacles, an ultrasonic sensor is chosen because this sensor measures the distances accurately. The sensor is placed 341
Fig. 3. Object detection and interpretation
IFAC PID 2018 342 Ghent, Belgium, May 9-11, 2018
Thi Thoa Mac et al. / IFAC PapersOnLine 51-4 (2018) 340–345
Fig. 4. Visualisation of LineMap formation
Fig. 6. The cost function graph of the shortest path finder method where G is the cost from the start node the present node. H is the heuristic that estimate the cost of the cheapest path from that node to the goal. The main difference is that shortest path finder does not use a grid but uses its start and goal location, and the end points of all obstacles to construct a graph in the map. The algorithm is presented in Fig 5. To show the concept, one example is depicted in Fig 6. From the start location, the possible paths are from start to node 1 or from start to node 2. The F -value from start to node 1 is calculated as the sum of its G-value (see Table 1), and the H-value of node 1 (as shown in Fig 6). Table 1. G-values for all possible edges Path start → 1 start → 2 1→2 2→3 2→4 2→5 2→6 3→4 3→5 3→6 4→5 6 → target
G - value 4,410 7,210 7,000 4,000 13,42 12,65 13,60 10,82 9,850 10,77 1,000 3,160
Table 2. The F cost value
Fig. 5. Shortest path finder algorithm flowchart. path finder method. In this section, we introduce about shortest path finder method which used in the leJOS programming language. Similar to the A* algorithm, the path is found by searching among all possible paths to the target the one that has smallest cost. The cost function of the shortest finder method is given in (1) F =G+H (1) 342
Path start start start start start start start start
→1 →2 →1→2 →2→3 →2→4 →2→5 →2→5 → 2 → 6 → target
F -value 23,82 23,97 28,38 25,13 24,75 23,98 23,97 23,97
As the F -value for the path between start and node 1, is lower than the one between start and node 2, node 1 is chosen for being added to the closed list. Next only one path is possible starting from node 1, namely the path to node 2. But similarly as explained for the A* algorithm,
IFAC PID 2018 Ghent, Belgium, May 9-11, 2018
qr
e −
Thi Thoa Mac et al. / IFAC PapersOnLine 51-4 (2018) 340–345
[vR , vL ]T PID Controller
Lego robot
343
q
qm Sensor feedbacks Fig. 7. Mobile robot control system.
Fig. 8. The Kinematic model of the Lego robot. the F -value when traveling directly from start to node 2 is lower than the one when traveling along the combined path start-node 1-node 2, as indicated in Table 2. So consequently node 1 is deleted of closed and is replaced by node 2. Next, four paths are possible for the robot to travel with node 2 as a starting point namely from node 2 to node 3, node 4, node 5 and to node 6. The F -values for these paths are again given in Table 2. As can be seen in the table, the next node that will added to the closed list will be node 6, as the F -value from node 2 to node 6 is the lowest of the four paths starting from node 2. Finally the only path possible starting from node 6 is the path to the target location, and with that making an end to the shortest path search. 4. MOBILE ROBOT CONTROL SYSTEM
Fig. 9. Scanning unknown obstacles with ultrasonic sensor x˙ r cos θ r cos θ y˙ = r sin θ r sin θ vR (3) vL r r −D θ˙ D 5. EXPERIMENTAL RESULTS
5.1 Sensor accuracy To decide which sensor to use to measure distances, an experiment has been performed. Both the ultrasonic sensor and the infrared sensor were tested. A comparison between real and measured values is shown in Table 3. It is clear that the ultrasonic sensor is the most accurate one: ± 1cm. The measured distances of the infrared sensor seems to be different depending on the material of the object measured. Therefore, the ultrasonic sensor is used for obstacle detection in this project. Table 3. The Distance Measurement
To this day, over 95 percentage of industrial applications are predominantly controlled by (PID) controllers due to is simplification and accuracy. A PID controller includes: Ki GP ID (s) = Kp + + Kd s (2) s where Kp , Ki , Kd denote the coefficients for the proportional, integral, and derivative terms. The proposed control system for the Lego robot in this study is shown in Fig 7. The control system consists of a PID controller, the kinematic model of mobile robot, a reference trajectory generator and two encoders which provides odometric information. In which vL and vR are the speed of the left and right wheel in cm/s respectively. Input, output, measured parameters are vectors: qr =[xr , yr , θr ]T , q =[x, y, θ]T , qm =[xm , ym , θm ]T . A schematic of the kinematic model is shown in Fig 8. The kinematic model is presented in equation (3). The parameters D and r are respectively the distance between two wheels and the wheel radius. 343
Real Distance 0 15 30 45 60 75 90 105 120
Infrared sensor black 0 18 34 44 51 ∞ ∞ ∞ ∞
Infrared sensor color 0 24 40 48 ∞ ∞ ∞ ∞ ∞
Ultrasonic Sensor 0 16.9 30.2 45.6 60.2 75.1 91.1 105.0 120.1
The algorithm used to detect unknown obstacles, is described in Section 2. In the lab-scale environment, three known obstacles were placed around the robot and their actual positions were compared to the positions measured by the robot’s sensor. Fig 10 shows the results of this test. These results are satisfactory, as the measured position is very close to the real obstacle.
IFAC PID 2018 344 Ghent, Belgium, May 9-11, 2018
Thi Thoa Mac et al. / IFAC PapersOnLine 51-4 (2018) 340–345
Fig. 11. Landing in presence of known/unknown obstacles results in the sequence of time 5.3 Autonomous navigation experiment In this experiments, Lego Mindstorms EV3 was moving in presence of uncertain obstacles. The yellow long wall is a known obstacle, while the red and blue walls are unknown obstacles. The start point and target point are marked on the map. Experiments were carried out to test the ability of the robot performing autonomous task. As soon as the robot detected unknown obstacle by ultrasonic sensor, the position of that obstacle is provided. The results show that the Lego robot avoid all obstacles on its path and stop on the target position. The intended accuracy has been achieved using an equipped ultrasonic sensor. 6. CONCLUSION
Fig. 10. (a.) Optimal path calculated by Shortest Path Finder method taking into account the robot size red path; (b.) compare with the path obtained by A* algorithm - pink path; (c.) Real path experiment by Shortest Path Finder method. 5.2 Shortest path finder experiment In this experiment, we evaluated the shortest path finder method by simulation and experiment. We also compared it with A* algorithm. Fig 10 presents the path obtained by shortest path finder algorithm - red path (a.); this path is compared to the one obtained by A* algorithm - pink path (b.) and the real path obtained by experiment - blue path (c.). The shortest path finder algorithm’s path is shorter than A* algorithm’s path. Moreover, the ShortestP athF inder only uses the endpoints of obstacles as nodes, it has a much shorter calculation time, as the amount of endpoints of obstacles is usually much smaller than the amount of cells in a grid used in A* algorithm. It is also able to pass closer to obstacles than A*, as A* uses the middle of a grid cell as a way-point. This is a reason the ShortestP athF inder algorithm is suggested in this project. Fig. 10 c shows that the robot is able to track the designed path accurately. 344
The design of automatics machine and robot is a discipline that needs practical work at the learning process. To carry out it at the laboratory with the students, the LEGO system is proposed. This platform provides basically 2 advantages: its low price and good features. This study proposed an effective educational hands-on testbed so that students can learn a concept and design process for using sensor system, path planning, and control processing. This aim of the added project is making an autonomous robot that is able to drive to a goal while avoiding known and unknown obstacles. In paper, we introduced ShortestP athF inder algorithm, which is a modified version of the A* algorithm. This comparison results showed the ShortestP athF inder algorithm to be the superior to the A* algorithm. The ultrasonic sensor was implemented to detect unknown obstacles, as it measures distances very accurately. Detecting unknown obstacles and determining their size was also done by using the ultrasonic sensor. The Lego robot was able to autonomously navigate in an unknown environment. The results of the experiments were satisfying. The achievements of our teaching method compared to classical robotic classes are: (i) the applications of a project-based learning approach as teaching methodology; (ii) the development of an effective real-time autonomous navigation approach for mobile robot. REFERENCES Rodriguez, C., Guzman, J. L., Berenguel, M., Dormido, S. (2016), Teaching real-time programming using mobile
IFAC PID 2018 Ghent, Belgium, May 9-11, 2018
Thi Thoa Mac et al. / IFAC PapersOnLine 51-4 (2018) 340–345
robots, IFAC-PapersOnLine vol. 49 (6), pp. 10-15. Kim, S., Oh, H., Kolaman, A., Tsourdos, A., White, B., Guterman, H. (2011), Educational hands-on test bed using Lego robot for learning guidance, navigation, and control, IFAC Proceedings Volumes, vol. 44(1), pp. 5170-5175. Grandi, R., Falconi, R., Melchiorri, (2014), Robotic Competitions: Teaching Robotics and Real-Time Programming with LEGO Mindstorms, IFAC Proceedings Volumes, vol. 47(3), pp. 10598-10603. Zdesar, A., Blazic, S., Klancar, G. (2017), Engineering Education in Wheeled Mobile Robotics, IFAC Proceedings Volumes, vol. 50(1), pp. 12173-12178. Chevalier, A., Copot, C., Ionescu, C., DeKeyser, R. (2017), A three-year feedback study of a remote laboratory used in control engineering studies, IEEE Transactions on Education, vol. 60(2), pp.127-133.
345
345
Copot, C., Ionescu, C., DeKeyser, R. (2017), Interdisciplinary project-based learning at master level: control of robotic mechatronic IFAC-PapersOnLine, vol. 49(6), pp. 314319. ICT and Mechatronics course content, Ghent University, https://studiegids.ugent.be/2016/EN/studiefiches/ E019330.pdf Mac, T. T., Copot, C., Tran, D .T., De Keyser, R. (2016), Heuristic approaches in robot path planning: a surveys, Autonomous and Robotics System, vol. 86, pp. 13-28. Mac, T. T., Copot, C., Hernandez, A., De Keyser, R. (2016), Improved potential field method for unknown obstacle avoidance using UAV in indoor environment, IEEE Intentional Symposium on Applied Machine Intelligence and Informatics, Herlany, Slovakia, pp. 345350.