Triple-Rrts for robot path planning based on narrow passage identification

Abstract

Random sampling-based methods are popularly used in path planning of robots with multiple DoFs, but their performance deteriorates sharply when the robot configuration space has narrow passages. This paper employs Randomized Star Builder to identify narrow passages in the working space, and configures extensions in the narrow passages. Based on this, a new Triple-RRTs (Rapidly-exploring Random Tree) method is proposed to ensure that the algorithms will quickly explore the configuration space and effectively find the connection in the narrow passages. Simulations of robot path planning in 2D and 3D space testify its validity.