BALANCING EXPLORATION AND EXPLOITATION USING PSO FOR A DISTRIBUTED MULTI-ROBOT COORDINATION

Coordination is critical for a multi-robot system to maximize the team performance under dynamic environment. The major challenge in designing a distributed coordination algorithm is to design each individual agent’s behaviors of exploration and exploitation so that an optimal global task performance can be achieved in a dynamic environment, where each agent has only partial and noisy information about the global task. To improve the robustness as well as the efficiency of a distributed multi-robot system, in this chapter, a biological inspired swarm intelligence based coordination algorithm is proposed, where a virtual pheromone mechanism is used as the message passing coordination scheme among the robots, and a novel utility function for each robot is defined as a function of virtual pheromone, target visibility, and target redundancy. To prevent all of robots converge to some local minimum, a balance between the exploration and exploitation is adjusted using the particle swarm optimization (PSO) method. Extensive simulation has been conducted on both self-developed JAVA based simulator and an embodied robot simulator. The simulation results concretely demonstrate the self-adaptive multi-robot system can evolve continuously with the biological concepts and mechanisms, adapting itself to ever-changing environment. I