基于个性的群体强化学习算法

Groups reinforcement learning algorithm Based on personality

本文以RoboCup为平台,提出群体强化学习算法,该算法将个性行为绑定到信息Agent上,让具有不同个性的Agent充当合适的角色,基于共同的目标,Agent可能产生共同的意图,规划出共同的行为,使追求个体目标与整体目标能合理协调,性能分析表明本文算法适用于动态、实时、有干扰、对抗的环境中。

This paper presents groups reinforcement learning algorithm to RoboCup as a platform,the algorithm acts will be bundled with personalized information agent,a different personality to the role of agent as appropriate,based on common objectives,common agent may have the intention of planning common the acts in which the pursuit of individual goals and overall objectives reasonable coordination,performance analysis show that the algorithm applied to dynamic,real-time,interference,a confrontational environment.