Constructing an Autonomous Agent with an Interdependent Heuristics

When we construct an agent by integrating modules, there appear troubles concerning the autonomy of the agent if we introduce a heuristics that dominates the whole agent. Thus, we design an agent that has an interdependent heuristics influenced by a module controlled by the heuristics, and we apply these agents into a problem of obtaining cooperation of Multi-Agents. We enable a present method that can solve the problem in a reinforcement learning context to be applied into a dynamic environment, and the improved method is embodied into the agent as the interdependent heuristics. We conduct experiments comparing the proposed agents with agents such as those ones each of which has a heuristics controlled by a supervisor, then we empirically confirm that the proposed agent having the interdependent heuristics is the most flexible of all the tested agent.