Improved RRT path planning algorithm based on growth evaluation

Abstract

Path planning algorithm is a key research problem in the application of autonomous mobile robots. RRT algorithm is one of the excellent methods for robot path planning. However, RRT algorithm has drawbacks such as high time consumption, high number of samples, low operational efficiency. On the basis of an adaptive resolution Octree Map, this paper presents an improved RRT path planning algorithm through establishing a growth point evaluation function. By adding the restriction through the evaluation function to random growing points of RRT, the random growth can be turned into purpose-oriented, and redundant growth points and routes can be eliminated. Besides, the growth path is reselected and rewired by the parent node and candidate nodes obtained through the growing process. Through reselection and rewiring, the route is continuously optimized and smoothed. Experiments are carried out for performance evaluation. Experiment results indicate that, comparing with the traditional RRT, RRT* and B-RRT*, the improved RRT algorithm can eliminate the redundant bifurcations of the growth tree reduce the number of sampling times, and greatly improves the growth efficiency.