Extended RRT-based path planning for flying robots in complex 3D environments with narrow passages

Abstract

Sampling-based methods such as Rapidly Exploring Random Trees (RRT) and Probabilistic Road Maps (PRM) have been recognized as effective tools to solve the path planning problem for both ground mobile robots and flying robots in high-dimensional configuration space. However, the efficiency of the RRT planner will be decreased in complex environments with narrow passages. This paper presents a multiple RRTs-based path planning framework to improve the above mentioned problem. The key ingredient of the framework is a hybrid sampling strategy which takes advantage of the Randomized Star Builder (RSB) and uniform sampling. The RSB method can efficiently recognize narrow passage regions while avoiding unnecessary samples in the corners and dead ends, and generates milestones for growing multiple local trees from narrow passages. Moreover, uniform sampling is used to generate global RRT trees in order to capture global connectivity. Simulation results of 3D flying robots demonstrate the effectiveness of the proposed method. Comparisons between the proposed method and other RRT-based planners are also presented.