State Space Construction for Cooperative Behavior Acquisition in the Environments Including Multiple Learning Robots

Abstract

This paper proposes a method that acquires cooperative behaviors based on the estimation of the state vectors. In order to acquire the cooperative behaviors in multi robots environments, each learning robot estimates the local predictive model between the learner and the other objects separately. Based on the local predictive models, robots learn the desired behaviors using reinforcement learning. The proposed method is applied to a soccer playing sit-uation, where a rolling ball and other moving robots are well modeled and the learner's behaviors are successfully acquired by the method. Computer simulations and real experiments are shown and a discussion is given.