影响图及其在Robocup中的应用

Influence Diagrams and Its Application in Robocup

影响图是决策问题的图形表示。它被认为是对动态贝叶斯网络的扩展,是在贝叶斯网络的基础上增加了决策结点和效用结点。探讨影响图在Robocup中的应用,用分层影响图表示了截球问题的模型,并给出一种实现行为动态调整的算法。仿真结果表明该算法可行的。

Influence Diagrams (IDs) is a graphical knowledge representation of decision problem. It may be viewed as an extension of Bayesian Networks (BNs), with additional node types for decision and utilities. It is discussed the basic concept of IDs and its application in Robocup. The model of intercept problem and a new arithmetic of real-time action adjustment are also presented. The results of simulation experiment indicate that the arithmetic is feasible.