Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Trajectory planning and collision avoidance for an autonomous racing car
Thesis (MEng)--Stellenbosch University, 2024.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | en_ZA en_ZA |
| Published: |
Stellenbosch : Stellenbosch University
2024
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867614025676750848 |
|---|---|
| access_status_str | Open Access |
| author | Keanly, Oran |
| author2 | Engelbrecht, Japie |
| author_browse | Engelbrecht, Japie Keanly, Oran |
| author_facet | Engelbrecht, Japie Keanly, Oran |
| author_sort | Keanly, Oran |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (MEng)--Stellenbosch University, 2024. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/130555 |
| institution | Stellenbosch University (South Africa) |
| language | en_ZA en_ZA |
| last_indexed | 2026-06-10T12:45:28.762Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2024 |
| publishDateRange | 2024 |
| publishDateSort | 2024 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/130555 Trajectory planning and collision avoidance for an autonomous racing car Keanly, Oran Engelbrecht, Japie Stellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering. Trajectories (Mechanics) Automated vehicles Automobiles -- Collision avoidance systems Dynamic bicycle model Collision detection (Computer animation) UCTD Thesis (MEng)--Stellenbosch University, 2024. ENGLISH ABSTRACT: The autonomous racing problem is to traverse a race track in the fastest time possible without any human intervention. However, there are also safety concerns, as an autonomous race car must be able to react to unexpected obstacles on the track and safely navigate around such obstacles and continue racing. This thesis develops an autonomous driving system that makes use of a hierarchical motion planner. The global motion planner functions offline and generates the optimal motion plan for the entire race track. The local collision avoidance planner functions online and generates a motion plan for a limited horizon ahead of the autonomous car when an obstacle is detected in the path of the vehicle. A dynamic bicycle model is presented and coordinated relative to a curvilinear track model. Then the global planner formulates the motion planning problem as an optimal control problem, using the dynamic bicycle model, to minimise the lap time of the car. A Stanley controller is used to execute the motion plan. A collision prediction system is developed which uses LiDAR scan measurements to identify and cluster objects within the track. The car is then propagated forward in time and if a collision is detected then the local planner is activated. The local planner uses a kinematic bicycle model which is presented and coordinated relative to a curvilinear track model. The local planner that is developed in this thesis is an extension of the global planner that formulates the path planning problem as an optimal control problem for a limited horizon. However, the plan is constrained to free space in the track and minimises the deviation from the global path as well as the lap time. The proposed autonomous racing system is tested on a variety of different tracks and track elements using the F1Tenth simulation environment. Tests were performed to analyse the planning, path-tracking, and collision-avoidance capabilities of the car. The simulation results show that the proposed autonomous racing system is able to generate and execute a minimum-time global racing plan, and is able to detect and avoid unexpected obstacles on the race track using a local collision avoidance plan. A collision avoidance success rate of 90% was achieved for single obstacles. However, the success rate reduces as the number of obstacles increases. AFRIKAANSE OPSOMMING:Die outonome resiesprobleem behels die voltooi van ’n resiesbaan in die kortste moontlike tyd sonder enige menslike intervensie. Veiligheid is egter ook belangrik, aangesien ’n outonome resiesmotor in staat moet wees om te reageer op onverwagte hindernisse op die baan en dit moet kan omseil en onverstoord aanhou met die resies. Hierdie tesis ontwerp ’n outonome bestuurstelsel wat gebruik maak van ’n hi¨erargiese bewegingsbeplanner. Die globale bewegingsbeplanner funksioneer aflyn en genereer die optimale bewegingsplan vir die volledige resiesbaan. Die plaaslike botsingsvermydingsbeplanner funksioneer in re¨ele tyd en genereer ’n bewegingsplan vir ’n beperkte horison voor die outonome voertuig wanneer ’n hindernis in die pad van die voertuig waargeneem word. ’n Dinamiese fietsmodel word voorgestel en geko¨ordineer relatief tot binne ’n kurwilineˆere baanmodel. Daarna formuleer die globale beplanner die bewegingsbeplanningsprobleem as ’n optimale beheerprobleem deur gebruik te maak van die dinamiese fietsmodel om die rondetyd van die motor te minimaliseer. ’n Stanley-beheerder word ge¨ımplementeer om die bewegingsplan uit te voer. ’n Botsingsvoorspellingsisteem is ontwikkel wat gebruik maak van LiDARskanderingmetings om voorwerpe op die baan te identifiseer en te groepeer. Die motor word dan voorwaarts in tyd gepropageer, en indien ’n botsing waargeneem word, word die plaaslike beplanner geaktiveer. Die plaaslike beplanner maak gebruik van ’n kinematiese fietsmodel wat voorgestel en geko¨ordineer word relatief tot die kurwilineˆere baanmodel. Die plaaslike beplanner wat in hierdie tesis ontwikkel is, is ’n uitbreiding van die globale beplanner wat die padbeplanningsprobleem formuleer as ’n optimale beheerprobleem vir ’n beperkte horison. Die beplanning is egter beperk tot die vrye spasie op die baan en minimeer afwyking van die globale pad, sowel as die rondetyd. Die voorgestelde outonome resiesstelsel is getoets op ’n verskeidenheid van uiteenlopende bane en baanelemente deur gebruik te maak van die F1Tenth-simulasie-omgewing. Toetse is uitgevoer om die beplanning, padvolging, en botsingsvermydingsvermo¨ens van die motor te analiseer. Die simulasie resultate wys dat die voorgestelde outonome resiesstelsel in staat is om minimum-tyd globale resiesplanne te genereer en uit te voer, en in staat is om onverwagte hindernisse op die resiesbaan waar te neem en te vermy deur gebruik te maak van ’n plaaslike botsingsvermyding plan. Die botsingsvermyding stelsel behaal ’n sukseskoers van 90% vir enkele hindernisse. Die sukseskoers daal egter soos wat die aantal hindernisse vermeerder. Masters 2024-03-04T08:41:38Z 2024-04-26T21:52:33Z 2024-03-04T08:41:38Z 2024-04-26T21:52:33Z 2024-03 Thesis https://scholar.sun.ac.za/handle/10019.1/130555 en_ZA en_ZA Stellenbosch University xiv, 128 pages : illustrations. application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Trajectories (Mechanics) Automated vehicles Automobiles -- Collision avoidance systems Dynamic bicycle model Collision detection (Computer animation) UCTD Keanly, Oran Trajectory planning and collision avoidance for an autonomous racing car |
| title | Trajectory planning and collision avoidance for an autonomous racing car |
| title_full | Trajectory planning and collision avoidance for an autonomous racing car |
| title_fullStr | Trajectory planning and collision avoidance for an autonomous racing car |
| title_full_unstemmed | Trajectory planning and collision avoidance for an autonomous racing car |
| title_short | Trajectory planning and collision avoidance for an autonomous racing car |
| title_sort | trajectory planning and collision avoidance for an autonomous racing car |
| topic | Trajectories (Mechanics) Automated vehicles Automobiles -- Collision avoidance systems Dynamic bicycle model Collision detection (Computer animation) UCTD |
| url | https://scholar.sun.ac.za/handle/10019.1/130555 |
| work_keys_str_mv | AT keanlyoran trajectoryplanningandcollisionavoidanceforanautonomousracingcar |