Collided Path Replanning in Dynamic Environments Using RRT and Cell Decomposition Algorithms

Abstract

The motion planning is an important part of robots’ models. It is responsible for robot’s movements. In this work, the cell decomposition algorithm is used to find a spatial path on preliminary static workspaces, and then, the rapidly exploring random tree algorithm (RRT) is used to validate this path on the actual workspace. Two methods have been proposed to enhance the omnidirectional robot’s navigation on partially changed workspace. First, the planner creates a RRT tree and biases its growth toward the path’s points in ordered form. The planner reduces the probability of choosing the next point when a collision is detected, which in turn increases the RRT’s expansion on the free space. The second method uses a straight planner to connect path’s points. If a collision is detected, the planner places RRTs on both sides of the collided segment. The proposed methods are compared with the others approaches, and the simulation shows better results in term of efficiency and completeness.