Cost-balanced cooperation protocol in multi-agent robotic systems

This paper proposes a cooperation protocol based on the cost-balanced strategy for the Object-Sorting Task in multi-agent robotic systems. The protocol coordinates agents for carrying objects to destinations efficiently and effectively. Each agent autonomously makes subjective optimal decision, then the coordination algorithm resolves their conflicts by balancing the load, which is measured in terms of cost. Since coordination can be performed simultaneously with agent movement, it incurs very little overhead. The protocol is efficient because every agent runs the same algorithm to obtain the common results without further communication. Implementation of the protocol is realized on a distributed modular agent architecture for design simplicity, flexibility, and reactivity. Experimental results have shown that (1) the protocol has better performance than a previously proposed help-based cooperation protocol, (2) the protocol is flexible, and (3) the protocol can effectively utilize the agent power to achieve linear and superlinear speedup in most cases.