Abstract: Estimating the action’s effect of agent in RoboCup simulator league based on Rough Set is discussed and when the action’s effect is not good, choosing one ancillary action which is good for it by adjusting the values of the controllable attributions is presented. It makes the agent can determine the next action effectively according to the current status in the field. Taking shooting for example, this paper certify that this approach is effective by testing.

Key words: rough set, RoboCup, action estimation, action choosing

摘要： 基于粗糙集的决策分析方法，对RoboCup仿真球队中Agent的行为执行效果进行评测，并且在当前行为的执行效果不理想的情况下，通过适当的调节可控属性值来选择有助于此行为的辅助行为。使得Agent能够根据当前场上的状态，更有效地决定下一步的行动。并以射门为例，通过实验证明了该方法的有效性。

关键词: 粗糙集, 机器人足球, 行为评测, 行为选择

CLC Number: 

• TP18