Jixie chuandong (Sep 2024)
Path Planning of Manipulators with the Improved RRT Algorithm in Complex Environment
Abstract
Aiming at the problems of the standard rapidly exploring random tree (RRT) algorithm in a complex environment, such as blind expansion, falling into local search, easy planning failure, low sampling success rate, and long paths, an adaptive goal-oriented strategy combined with an alternative strategy for regional sampling and an improved RRT algorithm of the greedy pruning strategy was proposed. Based on the kinematics of the manipulator, the envelope was used to simplify the manipulator model to improve the efficiency of collision detection. The adaptive goal-oriented strategy solved the problems of blind search, low search success rate, and difficult convergence of the RRT algorithm in complex environments; the regional sampling alternative strategy solved the problems of the RRT algorithm easily falling into local search, low sampling success rate, and long sampling time; the greedy pruning strategy eliminated redundant nodes and shortened the path, improved the path quality, and enhanced the robustness of the algorithm. In the Matlab and robot operating system (ROS), the obstacle avoidance simulation planning was carried out for different scenarios. The results show that the average search success rate of the improved RRT algorithm has increased by 82.4%, the average sampling success rate has increased by 67.5%, and the average path planning success rate has increased by 70%. The average time efficiency is increased by 81.9%, and the average path length is shortened by 63.05%. Finally, the practicability and effectiveness of the algorithm were further verified by the trajectory planning of the physical manipulator.
