Multiagent Learning of Coordination in Loosely Coupled
Chao Yu, Minjie Zhang, Fenghui Ren, Guozhen Tan · 2014
Multiagent learning is a promising technique for agents to learn efficient coordinated behaviors in multiagent system s. In multiagent learning, concurrent multiple distributed learning processes can make the learning environment non-stationary for each individual learner. Developing an efficient learning approach to coordinate age nts' behaviors in this dynamic environment is a difficult problem, especial ly when agents do not know the domain structure and have only local observability of the environment. In this paper, a coordinated multiagent learning approach is proposed to enable agents to learn efficient coor dinated behaviors by exploiting agent independence in loosely coupled multiagent systems. The main feature of the proposed approach is to explicitly quantify and dynamically adapt agent independence during learning so that agents can make a trade-off between a single-agent learning process and a coordinated learning process for an efficient decision making. The proposed approach is employed to solve two-robot navigation problems in different scales of domains. Experimental results show that agents using the proposed approach can learn to act in concert or independently in different areas of the environment, which results in great computational savings and near optimal performance.