【摘要】 强化学习在多Agent系统中面对的最大问题就是随着Agent数量的增加而导致的状态和动作空间的指数增长以及随之而来的缓慢的学习效率。采用了一种局部合作的Q-学习方法,只有在Agent之间有明确协作时才考察联合动作,否则,就只进行简单的个体Agent的Q-学习,从而使的学习时所要考察的状态动作对值大大减少。最后算法在捕食者-猎物的追逐问题和机器人足球仿真2D上的实验结果,与常用的多Agent强化学习技术相比有更好的效能。更多还原

【Abstract】 Reinforcement learning in Multi-Agent Systems suffers from the fact that both the state and the action space scale exponentially with the number of Agents,which also lead to low learning speed.In this paper,the authors investigate a regional cooperative of the Q-function by only considering the joint actions in those states in which coordination is actually requires.In all other states Single-Agent Q-learning is applies.This offers a compact state-action value representation,without compromising much in terms of solution quality.The authors have performed experiments in the predator-prey domain and robocup-simulation 2D which is the ideal testing platform of Multi-Agent Systems and compared this algorithm to other Multi-Agent reinforcement learning algorithms with promising results.更多还原