关键词: 渐进最优快速扩展随机树, 运动规划, 冗余机械臂, 启发式采样, 重规划

Abstract: Aiming at the problems of low precision, poor environmental adaptability of RRT*FN algorithm, an improved RRT*FN motion planning algorithm for manipulators was proposed. Firstly, in the iteration processes, combining the advantages of target biased random sampling and ellipsoid subset sampling, a new heuristic method was constructed to constrain the sampling areas, so as to ensure better search paths. Then, in expanding node processes, the preset values of total number of nodes in the tree were configured, and the leaf nodes in the tree were deleted by weighting method, which avoided the infinite growth of tree sizes. Finally, under the dynamic environment, a heuristic replanning method of node pruning and connection was adopted to effectively improve the adaptability of dynamic environment. Experimental results show that the algorithm herein has faster convergence speed and higher efficiency in planning processes, and has strong environmental adaptability.

Key words: rapidly-exploring random trees star fixed nodes(RRT*FN), motion planning, redundant manipulator, heuristic sampling, replanning