基于换位原理的对手建模模型  

A Model of Opponent Modeling Based on Transposition Principle

作　　者：顿文力[1] 孟庆春[1] 宋长虹[1] 张艳[1] 

机构地区：[1]中国海洋大学计算机科学系,山东青岛266071

出　　处：《中国海洋大学学报（自然科学版）》2004年第1期109-114,共6页Periodical of Ocean University of China

基　　金：山东省自然科学基金 ( Y2 0 0 2 G18)资助

摘　　要：对抗性多机器人系统是 1个典型的多智能体系统 (MAS)。在 MAS中 ,建立对手模型 ,并进行意图和规划识别是对抗活动的基本前提。BDI是描述思维的有效手段 ,可用来建立通用的对手思维状态模型。本文在 BDI模型的基础上结合换位原理 ,提出了 1种对抗性多机器人系统中的对手建模模型 ,并将此方法应用到足球机器人系统 ,获得了较好的效果。An adversarial multi-robot system is a typical multi-agent system (MAS). Developing an opponent model and recognizing opponent's intentions and plans are prerequisite to an MAS. BDI (Belief, Desire, Intention) is an effective means to describe a mental state, and it can be used to develop a generic opponent mental state model. In this paper, based on the model of BDI and combined with the transposition principle, a new method of modeling opponents in an adversarial multi-robot system is proposed. The method has been applied to a soccer robot system and the experimental results show that it can well enhance the system's performance and intelligence degree.

关 键 词：换位原理 对手建模 模型 多智能体系统 多机器人系统 意图预测 

分 类 号：TP424.6[自动化与计算机技术]