TUTCRIS - Tampereen teknillinen yliopisto

TUTCRIS

FORMI: A Fast Holonomic Path Planning and Obstacle Representation Method Based on Interval Analysis

Tutkimustuotosvertaisarvioitu

Yksityiskohdat

AlkuperäiskieliEnglanti
OtsikkoProceedings of the IEEE 2019 9th International Conference on Cybernetics and Intelligent Systems and Robotics, Automation and Mechatronics, CIS and RAM 2019
KustantajaIEEE
Sivut398-403
Sivumäärä6
ISBN (elektroninen)9781728134581
DOI - pysyväislinkit
TilaJulkaistu - 1 marraskuuta 2019
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisussa
TapahtumaIEEE International Conference on Cybernetics and Intelligent Systems and Robotics, Automation and Mechatronics - Bangkok, Thaimaa
Kesto: 18 marraskuuta 201920 marraskuuta 2019

Conference

ConferenceIEEE International Conference on Cybernetics and Intelligent Systems and Robotics, Automation and Mechatronics
MaaThaimaa
KaupunkiBangkok
Ajanjakso18/11/1920/11/19

Tiivistelmä

This paper presents a path planning approach for mobile robots in 2D spaces. The algorithm uses a quadtree decomposition where the discretization precision is improved until a path to the goal is found if one exists. The algorithm uses interval analysis-based methods to categorize the quadtree decomposition to occupied, free and partly occupied cells. The proposed algorithm is compared against other concurrent path planning algorithms, A on an ordinary quadtree, A for shortest path on a binary occupancy grid, and a Dijkstra's algorithm for lowest collision probability in a continuous-valued occupancy grid, in five different scenarios.Compared to the other methods, the main advantage of our method is achieving a compromise between driving distance, safety, and computation time. The proposed algorithm was found to require significantly fewer collision checks in all scenarios while providing sub-optimum results, based on the obstacle distance and path length criteria. The algorithm is suitable for further extension to include non-euclidean measures and for higher dimensions of configuration spaces. The proposed algorithm will be publicly available on GitHub repository.