Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Cooperative target following and collision avoidance for multiple unmanned aerial vehicles
Thesis (MEng)--Stellenbosch University, 2020.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | en_ZA |
| Published: |
Stellenbosch : Stellenbosch University
2020
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613921709391872 |
|---|---|
| access_status_str | Open Access |
| author | Ubbink, Johan Bernard |
| author2 | Engelbrecht, Jacobus Adriaan Albertus |
| author_browse | Engelbrecht, Jacobus Adriaan Albertus Ubbink, Johan Bernard |
| author_facet | Engelbrecht, Jacobus Adriaan Albertus Ubbink, Johan Bernard |
| author_sort | Ubbink, Johan Bernard |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (MEng)--Stellenbosch University, 2020. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/109251 |
| institution | Stellenbosch University (South Africa) |
| language | en_ZA |
| last_indexed | 2026-06-10T12:43:50.076Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2020 |
| publishDateRange | 2020 |
| publishDateSort | 2020 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/109251 Cooperative target following and collision avoidance for multiple unmanned aerial vehicles Ubbink, Johan Bernard Engelbrecht, Jacobus Adriaan Albertus Stellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering. Cooperative path planning Trajectory optimization Unmanned aerial vehicles Collision avoidance Model predictive control Robotics UCTD Thesis (MEng)--Stellenbosch University, 2020. ENGLISH ABSTRACT: In the field of robotics, there is an increasing number of applications for multiple Unmanned Aerial Vehicles (UAVs) to operate autonomously in a complex environment. This study contributes to the research field by presenting a trajectory planner for multiple rotary-wing UAVs to follow a moving ground vehicle cooperatively and collision-free through a three-dimensional domain with arbitrary static obstacles. An optimal control formulation is used to represent the target-following problem. The optimal control objective function is designed to minimise the position error in target following as well as the acceleration of the UAVs. Constraints are imposed to ensure that the trajectories are dynamically feasible and to avoid collisions between UAVs or between UAVs and obstacles. The optimal control problem is transcribed to a Nonlinear Program (NLP) and solved with the aid of a general optimisation solver. The optimisation solver starts with suitable initial estimates of the trajectories, which it iteratively improves until the optimal trajectories are obtained. These initial estimates are calculated with grid searches (based on the A* algorithm) that independently plan trajectories for all the UAVs. Each grid search neglects the other UAVs but does consider static obstacles in the environment. This decoupling (neglecting other UAVs) allows for initial estimates to be computed in parallel. The solver also requires that the objective function and constraint functions are smooth and differentiable. The static obstacles in the optimisation problem are represented by Euclidean Signed Distance Fields (ESDFs), which gives the distance to the nearest obstacle. The trajectory planner reacts to changes in the predicted target trajectory by using a replanning strategy similar to a Model Predictive Control (MPC) strategy. The replanning strategy continuously (at a fixed rate) plans for a receding horizon into the future. The trajectories are planned to ensure safe execution, even if an iteration of the replanning strategy fails. In-depth analysis of specific design parameters (planning resolution, planning horizon length and replanning rate) was performed, both from a theoretical perspective as well as simulation experiments. The analysis shows that some of the parameters are conflicting, and it is essential to balance the parameters to obtain a viable real-time implementation. The trajectory planner was combined with other components within the Robot Operating System (ROS), to form a target following system. Special care has been taken to ensure that the implementation could serve as a research platform for future projects in which multi-agent robotics systems can be developed and tested. The performance of the proposed trajectory planner is tested through simulation. First, examples are presented to illustrate the trajectory planner and target-following system. The trajectory planner is also tested in a large number of randomly generated scenarios, varying in complexity. The performance is analysed in terms of success rate, target following ability and planning time. The simulation results show that the UAVs can successfully follow a moving ground target while avoiding collisions with one another and with static obstacles for a large variety of targets and environments. AFRIKAANSE OPSOMMING: In die veld van robotika is daar ’n toenemende aantal toepassings vir verskeie onbemande vliegtuie om saam te werk op ’n gegewe opdrag. Hierdie studie dra by tot die navorsingsveld deur ’n trajekbeplanner te ontwerp vir verskeie rotor-vlerk vliegtuie om saam ’n teiken op die grond te agtervolg, terwyl hulle botsings met mekaar en die omgewing vermy. Die teikenvolging probleem word voorgestel as ’n optimale beheerprobleem. Die koste funksie is ontwerp om die volgingsfout sowel as onnodige versnelling deur die vliegtuie te minimeer. Die optimeringsprobleem word beperk om te verseker dat die trajekte uitvoerbaar is, sowel as om botsings tussen vliegtuie en botsings met die omgewing te vermy. Met die gebruik van trajekoptimeringstegnieke word die optimale beheerprobleem na ’n nie-lineˆere program (NLP) oorgeskryf en met behulp van ’n algemene optimerings oplosser opgelos. Die optimerings oplosser benodig ’n geskikte aanvanklike afskatting van die trajek, en die afgeleide van al die funksies moet bereken kan word. Die projek ondersoek die gebruik van ’n soekalgoritme om die aanvanklike skatting te gee. Die soekstrategie neem nie die ander vliegtuie in die omgewing in ag nie, maar vermy hindernisse in die omgewing. Deurdat die vliegtuie nie mekaar in ag neem nie, kan aanvanklike afskattings in parallel bereken word. Die statiese omgewing word voorgestel deur ’n veld wat die afstand tot die naaste hindernes aandui. Die projek pas die konsep van modelvoorspelling-beheerstrategie (MPC) toe om die trajekte intyds te beplan. Die MPC-strategie beplan voortdurend (teen ’n vaste tempo) vir ’n horison in die toekoms in. Die trajekte word so beplan dat dit veilige uitvoering verseker, selfs as ’n iterasie van die trajekbeplanner faal. ’n Analiese is gedoen om die inpak van die beplanningsresolusie, beplanningshorison en herbeplanningstempo te meet. Die resultate wys dat die veranderlikes in stryd is met mekaar, en versigtig gekies moet word om die trajekte intyds te beplan. Die beplanner is ge¨ımplementeer in die Robotic Operating System (ROS) en getoets in samewerking met ’n trajek uitvoerder in ’n Gazebo-simulasie. Afgesien van die spesifieke toepassings wat uiteengesit en getoets is, is daar veral gesorg dat die implementering kan dien as ’n navorsingsplatform vir toekomstige projekte waarin robotiese stelsels ontwikkel en getoets kan word. Verskeie gevallestudies word aangebied om spesifieke kenmerke van die trajekbeplanner uit te lig en om die prestasie te evalueer. Die resultate toon dat die beplanner in staat is om botsingsvrye trajekte vir ’n verskeidenheid teikens en omgewings te genereer. Masters 2020-11-30T19:50:59Z 2021-01-31T19:41:13Z 2020-11-30T19:50:59Z 2021-01-31T19:41:13Z 2020-12 Thesis http://hdl.handle.net/10019.1/109251 en_ZA Stellenbosch University 129 pages : illustrations application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Cooperative path planning Trajectory optimization Unmanned aerial vehicles Collision avoidance Model predictive control Robotics UCTD Ubbink, Johan Bernard Cooperative target following and collision avoidance for multiple unmanned aerial vehicles |
| title | Cooperative target following and collision avoidance for multiple unmanned aerial vehicles |
| title_full | Cooperative target following and collision avoidance for multiple unmanned aerial vehicles |
| title_fullStr | Cooperative target following and collision avoidance for multiple unmanned aerial vehicles |
| title_full_unstemmed | Cooperative target following and collision avoidance for multiple unmanned aerial vehicles |
| title_short | Cooperative target following and collision avoidance for multiple unmanned aerial vehicles |
| title_sort | cooperative target following and collision avoidance for multiple unmanned aerial vehicles |
| topic | Cooperative path planning Trajectory optimization Unmanned aerial vehicles Collision avoidance Model predictive control Robotics UCTD |
| url | http://hdl.handle.net/10019.1/109251 |
| work_keys_str_mv | AT ubbinkjohanbernard cooperativetargetfollowingandcollisionavoidanceformultipleunmannedaerialvehicles |