Distributed formation control in cluttered environments

Abstract

A graph theory based control mechanism that enables groups of ground moving nonholonomic robots is proposed. The mechanism allows the robot to dynamically manage formation shapes and follow the leader through environments with obstacles. It improves upon a state of the art formation control algorithm where a formation can be maintained without the need of inter-robot communications. Obstacle avoidance is designed to be scalable and allows the robots to dynamically manage their formation according to the environment. The formation is also capable of rebuilding itself when individual robots within the formation fail. The algorithm has been tested on a nonholonomic multi-robot system, with results showing that the proposed algorithm enables a formation to complete an obstacle course and regenerate original formation shapes within 12 seconds with no collisions.