A new path planning method based on concave polygon convex decomposition and artificial bee colony algorithm. (7th January 2020)

• Record Type: Journal Article
• Authors: Li, Zhaoying; Zhang, Zhao; Liu, Hao; Yang, Liang
• Is Part Of: International journal of advanced robotic systems. Volume 17:Number 1(2020:Jan./Feb.)
• Publication Date: 2020-01-07
• Subjects: Path planning -- free space -- artificial bee colony (ABC) -- map decomposition; Robotics -- Periodicals; Robotics; Periodicals; 629.892
• DOI: 10.1177/1729881419894787
• Languages: English
• ISSNs: 1729-8806

Abstract

Free space algorithms are kind of graphics-based methods for path planning. With previously known map information, graphics-based methods have high computational efficiency in providing a feasible path. However, the existing free space algorithms do not guarantee the global optimality because they always search in one connected domain but not all the possible connected domains. To overcome this drawback, this article presents an improved free space algorithm based on map decomposition with multiple connected domains and artificial bee colony algorithm. First, a decomposition algorithm of single-connected concave polygon is introduced based on the principle of concave polygon convex decomposition. Any map without obstacle is taken as single-connected concave polygon (the convex polygon map can be seen as already decomposed and will not be discussed here). Single concave polygon can be decomposed into convex polygons by connecting concave points with their visible vertex. Second, decomposition algorithm for multi-connected concave polygon (any map with obstacles) is designed. It can be converted into single-connected concave polygon by excluding obstacles using virtual links. The map can be decomposed into several convex polygons which form multiple connected domains. Third, artificial bee colony algorithm is used to search the optimal path in all the connected domains so as to avoid falling into the local minimum. Numerical simulations and comparisons with existing free space algorithm and rapidly exploring random tree star algorithm are carried out to evaluate the performance of the proposed method. The results show that this method is able to find the optimal path with high computational efficiency and accuracy. It has advantages especially for complex maps. Furthermore, parameter sensitivity analysis is provided and the suggested values for parameters are given.