Abstract | ||
---|---|---|
The problem of 3D path planning has always been important and challenging in the development of automatic vehicles. In order to achieve a fast 3D path planning of high quality, a novel differential evolution (DE) with the aid of a heuristic procedure, i.e., HeuDE, is proposed in this paper. HeuDE is composed by an initialization phase and an evolution phase. In the initialization phase, the heuristic procedure is responsible to search for a potential problem space such that the differential evolution algorithm can quickly find a feasible and high-quality path in the subsequent evolution phase. The heuristic procedure works by constructing potential paths based on the available heuristic information extracted from a cube-based 3D modeling. To utilize the heuristic information, two strategies for waypoint selection are developed for the step-by-step path construction in the heuristic procedure. Experimental results demonstrate the good performance of the proposed HeuDE for 3D path planning and verify that the combination of the heuristic procedure with DE is mutually beneficial. Further experiments on HeuDE of a smaller population size prove its ability for fast 3D path planning. |
Year | DOI | Venue |
---|---|---|
2017 | 10.1145/3067695.3076013 | GECCO (Companion) |
Keywords | Field | DocType |
3D path planning, differential evolution, heuristic information | Motion planning,Heuristic,Any-angle path planning,Mathematical optimization,Computer science,Differential evolution,Artificial intelligence,Initialization,Null-move heuristic,Machine learning,Consistent heuristic,Fast path | Conference |
ISBN | Citations | PageRank |
978-1-4503-4939-0 | 1 | 0.35 |
References | Authors | |
10 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ning Ma | 1 | 45 | 5.95 |
Xue Yu | 2 | 17 | 3.53 |
Wei-Neng Chen | 3 | 143 | 13.16 |
Jun Zhang | 4 | 326 | 21.82 |