Reinforcement Learning with Symbiotic Relationships for Multiagent Environments
Shingo Mabu, Masanao Obayashi, Takashi Kuremoto · Journal of Robotics Networking and Artificial Life · 2015
Multiagent systems, where many agents work together to achieve their objectives, and cooperative behaviors between agents need to be realized, have been widely studied In this paper, a new reinforcement learning framework considering the concept of "Symbiosis" in order to represent complicated relationships between agents and analyze the emerging behavior is proposed.In addition, distributed state-action value tables are designed to efficiently solve the multiagent problems with large number of state-action pairs.From the simulation results, it is clarified that the proposed method shows better performance comparing to the conventional reinforcement learning without considering symbiosis.