Research on manipulator motion planning based on improved RRT algorithm

Abstract:
In the process of manipulator motion planning, the traditional RRT algorithm has problems such as blindness and randomness, and there will be some problems, which will lead to tortuous and lengthy planned paths, and waste computing time for collision detection of a large number of invalid nodes. Therefore, an improved algorithm based on the RRT algorithm is proposed in this paper. Firstly, at the starting point, optimize the sampling space and change the sampling method to avoid the regression of subsequent sampling nodes. Then, at the target point, improve the judgment conditions of the target, dynamically adjust the expanding range of the target point, and speed up the convergence of the node tree. In addition, in the process of node sampling planning, the bias factor of the target is introduced to construct the sampling function, which makes the path heuristic and directional in the process of finding. Finally, by constructing a simple environment and complex environment for simulation experiments, the improved algorithm is compared with the traditional RRT algorithm and RRT star algorithm. Here we verify that the pathfinding time and the number of nodes are relatively reduced by 73.6%, 92.6%, 54.6%, and 75.9%, which proves that the proposed algorithm has higher planning efficiency.