【摘要】 本文提出了一种基于强化学习的足球机器人带球任务的解决方案,带球方从阶段的开始到结束均尽力去控制足球,对方的队员则尽力从控球方抢夺足球。在本论文中控球方采用SARSA控制算法,防守方采用传统的策略。将这两种策略应用于4V3机器人比赛环境中,实验结果显示,经过多次的学习,在机器人的带球时间上取得了理想的效果。更多还原

【Abstract】 This paper propose a reinforcement learning solution to the soccer agent dribbling task,the controler try to control the soccer from the beginning to the end of a region,as an adversary attempts to gain possession.In this paper,the team possession use SARSA control algorithm,the adversary use the traditional strategy.These two strategies are applied to 4V3 robot competition environment.The result of experiments show that,after training period,we obtained the ideal effect of the robot dribbling task.更多还原