Multi-Agent Coordination Using Limit Cycles in Dynamic Environment

Abstract

The paper presents a novel neural oscillator-based limit cycle approach for multiple robots formation and navigation. A group of point-robots is directed to move towards a target as well as to form a circular shape surrounding it. Initially the algorithm for a stationary target in absence of obstacle is developed, and then it is extended to dynamic target and further to an obstacle avoidance including static and dynamic obstacles. An attraction property of neural oscillator-based limit cycle is used for the robots to move towards circle around target with an arbitrary radius. The group of pursuing robots reaches the target in such a way that they remain at a desired distance from it keeping equally distributed angle around it. Another finding is that a change of communication (or sensor) matrix makes a transient formation to be different, though the final shape remains the same. Finally, the effectiveness of the proposed method is verified through several simulations.