Improvement of path planning algorithm based on RRT

Abstract

In order to solve the problems of strong sampling randomness, slow convergence speed, long and tortuous path and poor adaptability of narrow channel in the path planning of robotic arm by using the rapid exploration random tree (RRT) algorithm, a hierarchical adaptive smooth pruning RRT* (HASP-RRT*) algorithm is proposed in this paper. Firstly, The sampling strategy of Gaussian distribution is introduced to focus the sampling on the key areas and reduce the frequency of invalid sampling. The angle constraint mechanism is designed to limit the direction of node expansion, and the decision function is used to avoid the algorithm falling into local extremum. Then, a multi-factor adaptive step-size strategy integrating goal orientation, obstacle safety distance and exploration progress is proposed to improve the expansion efficiency of the algorithm in complex environments. Finally, the triangle clipping interpolation is used to remove the redundant nodes of the path, and the non-uniform third-order B-spline is used to smooth the path to generate an optimized trajectory with continuous curvature. In order to verify the advantages of HASP-RRT algorithm, the experimental environment is constructed in MATLAB. By comparing with RRT algorithm, RRT* algorithm, G-RRT algorithm and informed-RRT* algorithm, the results show that compared with RRT algorithm, the convergence time of HASP-RRT algorithm is reduced by 89.5% on average, and the number of sampling nodes is reduced by 82.5% on average. Compared with the RRT* algorithm, the convergence time is reduced by 90% on average, and the number of sampling nodes is reduced by 82.5% on average. Compared with G-RRT* algorithm, the convergence time is reduced by 78.5% on average, and the number of sampling nodes is reduced by 74.5% on average. Compared with the informed- RRT* algorithm, the convergence time is reduced by 74.5% on average, and the number of sampling nodes is reduced by 73% on average. The simulation verification based on the UR5e manipulator further shows that the algorithm can effectively suppress the sudden change of the joint angle, generate a smooth and stable motion trajectory, and significantly improve the efficiency and quality of the path planning of the manipulator.