Cooperative multi-agent transport based on coevolution

Achieving cooperative behavior is on of the central issues in multi-agent systems. A novel cooperative algorithm is proposed for multi-agent transporting multiple objects concurrently. The multi-agent system with decentralized architecture consists of several homogeneous agents, a communication scheme combining blackboard and point to point approach is used to exchange cooperation information among agents. Cooperative algorithm consists of task balance and agent-object sequences co-evolution. Simulation experiments that three agents transport eight objects were conducted for evaluating the presented approach. Results show that transportation task sequence can be optimized, and task can be performed with less time than previous method.