Adaptive and cooperative learning for Robocup agents
Jong-Yih Kuo, Frank Hsieh · 2008
In this paper, we study several adaptive learning strategies for robot agents in a Robocop game. A Q-learning based method is introduced to learning the mapping among agent’s actions. We apply these strategies to improve robot’s plan. In order to facilitate the development of shred understanding among game strategies, Piget’s cognitive theory is applied to the use of cooperative learning. This paper uses a RoboCup game to explain our approach.