An improved RRT* algorithm for mobile robots path planning

Abstract

Rapidly-exploring Random Tree Star(RRT*) is a path-planning algorithm based on Rapidly-exploring Random Tree(RRT). RRT* is applicable to complex and high-dimensional problems. Although RRT* algorithm is asymptotically optimal, its slow convergence rate makes it less efficient. To address this problem, an efficient optimal path-planning algorithm based on RRT* is proposed in this paper, which combines the advantages of Tropistic RRT* and Quick-RRT*. Firstly, an adaptive sampling strategy based on the tree growth is proposed to search an initial path in restricted space. Secondly, the path is optimized by node rejection after initial path is found. Finally, the improved algorithm expands the elements of possible parent vertices in ChooseParent and Rewire procedures. The simulation results demonstrate the feasibility of the proposed algorithm.