Cooperative navigation for multiple autonomous ground vehicles (AGVs) with kinematic constraints
| dc.contributor.advisor | Engelbrecht, Jacobus Adriaan Albertus | en_ZA |
| dc.contributor.advisor | Engelbrecht, Herman | en_ZA |
| dc.contributor.author | Viljoen, Ruan Matthys | en_ZA |
| dc.contributor.other | Stellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering. | en_ZA |
| dc.date.accessioned | 2020-11-30T20:56:23Z | |
| dc.date.accessioned | 2021-01-31T19:44:39Z | |
| dc.date.available | 2020-11-30T20:56:23Z | |
| dc.date.available | 2021-01-31T19:44:39Z | |
| dc.date.issued | 2020-12 | |
| dc.description | Thesis (MEng)--Stellenbosch University, 2020. | en_ZA |
| dc.description.abstract | ENGLISH ABSTRACT: This thesis presents the development of a system for the cooperative navigation of severalAutonomous Ground Vehicles (AGVs) within the same environment. A high-level systemarchitecture is designed that includes the following modular components: a cooperativetrajectory planner, a trajectory tracker, and a velocity controller. The cooperativetrajectory planner forms the highest level subsystem, and is responsible for finding collision-free trajectories for each vehicle. It does this using a decentralised coordination strategy,allowing for a more distributive and resilient system. The planning is accomplished foreach vehicle through the use of the Windowed Hierarchical Cooperative A* (WHCA*)multi-agent planning algorithm, modified so as to adhere to the kinematic constraintsof the vehicles. The second subsystem is the trajectory tracking module, which uses aModel Predictive Control (MPC) strategy to control the vehicles to track the plannedtrajectories, while also taking the kinematic constraints of the vehicle into account.Each of the subsystems were developed and tested using a simulation environmentmade with the ROS and Gazebo toolchain. This simulation environment was also used totest the overall performance of the integrated system. These tests were repeated using apractical setup with physical vehicles, so as to evaluate the performance of the system in areal world environment. In order to perform the practical tests, both the physical vehiclesand a vehicle pose estimation system were designed and built. The purpose of the vehiclepose estimation system was to find and track the pose of the vehicles, which was requiredby both the trajectory planning and tracking algorithms. The vehicle pose estimationwas accomplished through the use of the ArUco fiducial marker detection computer visionalgorithm.Both the simulation and practical tests show that the cooperative navigation algorithmswere able to successfully plan and execute trajectories using a decentralised coordinationstrategy, resulting in collision free navigation for all the vehicles involved. Both thetrajectory planning and the trajectory optimisation were able to execute within theirallowed time frame, which means the cooperative navigation system is viable for real-timeoperation. | en_ZA |
| dc.description.abstract | AFRIKAANSE OPSOMMING: Hierdie tesis beskryf die ontwerp van ’n stelsel wat gebruik kan word vir die gedesentrali-seerde samewerkingsnavigasie van verskeie outonome grondvoertuie binne dieselfde omge-wing. ’n Stelselargitektuur op ho ̈e vlak is ontwerp wat die volgende modulˆere komponentebevat: ’n ko ̈operatiewe trajekbeplanner, ’n trajekuitvoerder, en ’n snelheidskontroleerder.Die ko ̈operatiewe trajekbeplanner vorm die hoogste stelsel en is verantwoordelik vir dievind van botsingsvrye trajekte vir elke voertuig. Dit word gedoen met behulp van ’ngedesentraliseerde ko ̈ordineringstrategie, wat ’n meer verspreidende en betroubare stelselmoontlik maak. Die beplanning word vir elke voertuig gedoen deur gebruik te maak vandie Windowed Hierarchical Cooperative A* (WHCA*) veelagent beplanningsalgoritme,aangepas om te voldoen aan die kinematiese beperkings van die voertuie. Die tweedesubstelsel is die trajekuitvoeringmodule, wat gebruik maak van ’n Model Predictive Control(MPC) strategie om die betroubare uitvoering van die beplande trajekte te verseker, terwyldie kinematiese beperkings van die voertuig ook in ag geneem word.Elk van die substelsels is ontwikkel en getoets met behulp van ’n simulasie-omgewingwat gemaak is met die ROS en Gazebo gereedskapsketting. Hierdie simulasie-omgewingis ook gebruik om die algehele optrede te toets sodra al die substelsels in ’n holistieseoplossing ge ̈ıntegreer is. Hierdie toetse is herhaal met behulp van ’n praktiese opstellingmet fisiese voertuie om die optrede van die stelsel in ’n werklike wˆereldomgewing te evalueer. Om die praktiese toetse uit te voer, moes beide die fisiese voertuie en ’nvoertuigopsporingstelsel ontwerp en gebou word. Die doel van die voertuigopsporingstelselwas om die geskatte posisie van die voertuig te verskaf, wat deur die trajekbeplanning entrajekuitvoering algoritmes vereis word. Die voertuigopsporing is gedoen deur die ArUcomerker rekenaarvisie-algoritme te gebruik.Beide die simulasie en praktiese toetse toon dat die ko ̈operatiewe navigasie-algoritmesin staat was om trajekte suksesvol te beplan en uit te voer met behulp van die gedesentra-liseerde ko ̈ordineringstrategie, wat gelei het tot botsingsvrye navigasie vir al die betrokkevoertuie. Beide die trajekbeplanning en die trajekoptimalisering kon binne hul toegelatetydsbestek uitgevoer word, wat beteken dat die ko ̈operatiewe navigasiestelsel gebruikbaaris vir intydse werking. | af_ZA |
| dc.description.version | Masters | en_ZA |
| dc.format.extent | 166 pages | en_ZA |
| dc.identifier.uri | http://hdl.handle.net/10019.1/109322 | |
| dc.language.iso | en_ZA | en_ZA |
| dc.publisher | Stellenbosch : Stellenbosch University | en_ZA |
| dc.rights.holder | Stellenbosch University | en_ZA |
| dc.subject | Cooperative navigation | en_ZA |
| dc.subject | Autonomous ground vehicles | en_ZA |
| dc.subject | Kinematics -- Constraints | en_ZA |
| dc.subject | Model based predictive control | en_ZA |
| dc.subject | Trajectories (Mechanics) | en_ZA |
| dc.subject | Collision avoidance systems | en_ZA |
| dc.subject | UCTD | en_ZA |
| dc.title | Cooperative navigation for multiple autonomous ground vehicles (AGVs) with kinematic constraints | en_ZA |
| dc.type | Thesis | en_ZA |