Motion Planning of Redundant Dual-arm Based on Improved RRTstar

In view of the lack of search orientation and slow convergence speed of RRT * algorithm, this paper proposes an improved RRT * path planning algorithm. Firstly, after sampling random points, the attractive function is introduced to guide the expansion direction of the random tree. Then, the variable multiple step size method is used. Every time the random tree is expanded, if no collision is detected, the step size will be doubled in the next expansion, and so on; if collision is detected, the step size will be adjusted back to single step size for expansion, so as to speed up the algorithm and improve the efficiency. Then, after analyzing the workspace of the redundant double manipulator, the possibility of collision between the two arms is confirmed, and the loose coordination method of master-slave programming is used for the experiment. Simulation and experimental results show that compared with the original algorithm, the improved algorithm saves about 23% in time and 11% in path cost, which proves the effectiveness and efficiency of the proposed algorithm.