为了更好地解决一类特殊的Agent决策问题,提出行动驱动的马尔可夫决策过程的概念并分析了其理论模型.另外,提出行动驱动的马尔可夫决策过程相关问题的求解算法,并在RoboCup仿真2D比赛的不离身带球问题中对算法进行了实验.实验结果表明,新算法使Agent的带球性能有了较大的提高.新算法已经用于中国科大“蓝鹰”仿真2D机器人足球队,并在比赛中取得了较好的效果. In order to solve a special kind of problem of Agent decision-making, this paper proposes the concept of action-driven Markov decision process and analyzes its theoretical model.In addition, it proposes a solution algorithm of action-driven Markov decision process related problems, The experimental results show that the new algorithm has greatly improved the performance of the agent’s ball-in-ball. The new algorithm has been used to simulate the “Lanying” 2D robot soccer team, and achieved good results in the competition.