Multiagent Reinforcement Learning with Regret Matching for Robot Soccer

Figure 1

Architecture of MARL based on Markov game.