Distributed Control for Multirobot Systems with Collision-Free Motion Coordination

Abstract

This paper presents a distributed framework that ensures reliably collision-free behaviors in large-scale multirobot systems with switching interaction topologies. We show how a multirobot system achieves a natural collisionless aggregation through leader switching in the presence of obstacles. Although this method is mainly designed for collision avoidance, the velocity matching of robots is still achieved after the obstacle avoidance. The Separation (avoid robot-torobot collisions) rule is actively controlled through an artificial potential function of inter-robot distance. Based on the fluid theory, the Avoidance (avoid obstacles) rule is achieved by making robots track a set of generated virtual agents along the streamlines. In addition, a new species of "filtering agents" are designed to decrease the adverse effect. Simulation results are presented to demonstrate the effectiveness of our method.