RoboCup防守策略研究

Research on the RoboCup defensive strategy

Robocup仿真赛是研究 Agents 之间的对坑和协作理论比较理想的平台.球队的高度是由 Robocup的防守来决定,如何提高 Robocup的防守是一个很有挑战的问题.为了更好的理解和训练我把比赛中的 Agent 分解成多个一对一的任务.采用基于改进行的 Q算法来解决比赛中的防守问题.将该算法应用到广东工业大学的球队上,在实际比赛中取得了很好的效果.验证了基于改进的 Q算法在防守起到比较理想的作用.

Robocup soccer simulation is an excellent platform in which collaboration and counterwork among multi-agent are studied. Theheight of the team is decided by the Robocup defensive. How to improve the Robocup defense is very challenging problem. I takethe game in order to better understand and training Agent is decomposed into multiple tasks one-on-one. Based on improving theQ algorithm I am learning to solve the defensive problems in the game. Applying the algorithm to the team of Guangdonguniversity of technology, good results have been achieved in the actual game, Based on the improved Q learning is verified on thedefensive works as an ideal