Modular Learning System and Scheduling for Behavior Acquisition in Multi-agent Environment

Abstract

The existing reinforcement learning approaches have been suffering from the policy alternation of others in multiagent dynamic environments such as RoboCup competitions since other agent behaviors may cause sudden changes of state transition probabilities of which constancy is necessary for the learning to converge. A modular learning approach would be able to solve this problem if a learning agent can assign each module to one situation in which the module can regard the state transition probabilities as constant. This paper presents a method of modular learning in a multiagent environment, by which the learning agent can adapt its behaviors to the situations as results of the other agent's behaviors. Scheduling for learning is introduced to avoid the complexity in autonomous situation assignment.