Adaptive GB_RRT* Manipulator Path Planning Based on Grid Space

Abstract

To make the manipulator quickly plans a better obstacle avoidance path under different risk assembly,a rapidly exploring random tree star algorithm for adaptive goal bias based on grid space (SAGB_RRT*) was proposed.The Goal Bias Rapidly exploring Random Tree algorithm (GB_RRT) was optimized according to Rapidly exploring Random Tree star (RRT*) asymptotic optimization to make the search path converge towards the optimal solution.As to the issue that developed GB_RRT* required a large amount of calculation to traverse and search adjacent nodes in the joint space,grids was adopted to quickly find adjacent nodes and storage tree node for speeding up the algorithm calculation.Openlist was utilized to address the problem of repeated bias caused by target bias sampling strategy.In view of the difficulty of determining the target bias threshold in different environments,an adaptive target bias method was proposed,which enabled the algorithm to change growth strategy in real time according to the openlist feedback and hence reduce the algorithm Invalid extension,search time and path cost.To further reduce the path twists and costs,the greedy algorithm with variable interval was used to quickly optimize the path planned by SAGB_RRT* within limited time.In the simulation experiment,the proposed method was applied to the path planning of manipulators in different complex environments.The experimental results showed that the proposed algorithm could effectively decrease the search time and path cost and improve the planning stability.