A cooperation model using reinforcement learning for multi-agent

In multi-agent systems, the common goals of each agent are established and the problems are solved through cooperation and control among agents. Because each agent performs parallel processes in a multi-agent system, this approach can be easily applied to problems requiring parallel processing. The parallel processing prevents system performance degradation due to local error operation in the system. It also can reduce the solution time when the problem is divided into several sub-problems. In this case, each agent is designed independently providing a relatively simple programming model for solution of the problem. Further, the system can be easily expanded by adding new function agents. In the study of multi-agent systems, the main research topic is the coordination and cooperation among agents.