| Full text | |
| Author(s): |
Pinheiro, Felipe C. R.
;
Maximo, Marcos R. O. A.
;
Yoneyama, Takashi
;
Goncalves, LMG
;
Drews, PLJ
;
DaSilva, BMF
;
DosSantos, DH
;
DeMelo, JCP
;
Curvelo, CDF
;
Fabro, JA
Total Authors: 10
|
| Document type: | Journal article |
| Source: | 2020 XVIII LATIN AMERICAN ROBOTICS SYMPOSIUM, 2020 XII BRAZILIAN SYMPOSIUM ON ROBOTICS AND 2020 XI WORKSHOP OF ROBOTICS IN EDUCATION (LARS-SBR-WRE 2020); v. N/A, p. 6-pg., 2020-01-01. |
| Abstract | |
The path planning problem is an important issue in robotics. In the context of the Small Size Soccer league, a widely used algorithm for path planning is the Rapidly-Exploring Random Tree (RRT), which is a single-query planner. However, properties such as asymptotic optimality and anytime resolution may not be achieved by some of the algorithms used in the competition. Some of the possible path planning solvers with some of these properties are the Batch Informed Trees (BIT*), Fast Marching Trees (FMT*), and other RRT implementations (such as Extended Execution RRT - ERRT and RRT*). The contribution of this paper is to carry out a benchmark comparative investigation with a selected set of path planners found in the literature, evaluating their performance in the context of Robocup Small Size League Competition. The results show good convergence properties of BIT* when the planning time requirement is larger. ERRT performs well to find a solution, but it falls short concerning the solution quality. RRT* did not have a good convergence to optimal paths, but succeeded to find paths with short planning time. Since FMT* does not present anytime resolution, the algorithm did not present a satisfactory convergence with larger planning time. To improve the performance of the planners, it may be interesting to conceive heuristic strategies to allow trivial problems to be solved faster. (AU) | |
| FAPESP's process: | 20/04561-0 - Path planning for small size robots |
| Grantee: | Felipe Celso Reis Pinheiro |
| Support Opportunities: | Scholarships in Brazil - Master |