仮想ポテンシャル関数を用いたエージェントの相互制御方法に関する検討

抄録

One of the major problems of the technique that controls actions of agents by potential functions is to determine the function shapes suitable for the environment. The authors have proposed a method for autonomous acquisition of the parameters of potential functions by the agents with reinforcement learning, especially classifier systems, in multi-agent environment.
In this paper, we study a method to introduce temporary potential functions, which agents can set at desirable positions, so as to improve controllability of other agents. Experiments with a pursuit game are done, and the following are confirmed: acquisition of the temporary potential parameters for capturing the prey agent is possible; acquisition of the temporary potential parameters for the hunter agents to control each other and cooperatively capture the prey agent is also possible; the temporary potential increases the capturing probability of the prey agent.