基于情感计算和Q-learning的agent自主追逐行为过程研究  

Research of autonomous behavior of agent chasing based on affective computing and Q-learning

作　　者：李木军[1] 刘箴[1] 林君焕[1] 于力鹏[1] 

机构地区：[1]宁波大学信息科学与工程学院,浙江宁波315211

出　　处：《计算机应用研究》2014年第6期1710-1713,1718,共5页Application Research of Computers

摘　　要：针对目前智能体间追逐过程中对智能体的情感因素考虑不充分的问题,提出一种新的解决方案:首先通过情感建模将个性、情感融入以两个智能体为基元的追逐行为中,使其运动更有多样性;其次通过博弈论引导决策的选取;最后收集对方运动的轨迹点,用Q-learning加强学习方式学习归纳,以寻找最优追逐运动路径。在Visual Studio 2012编译环境下得到整个具有可信度的运动动画以及智能体的情感、体力等因素的变化规律图像。演示结果表明,此解决方案对于智能体间高效的追逐有很好的促进作用。At present,there is a problem in researches of multi-agents chasing behavior,which is that we considered affective factors of the agents inadequately.This paper presentsed a new solution.Firstly,it set up the model of personality and affecrive for chasing behavior of multi-agents,to make their motion more diversity.Then it guided their selection of decision making by game theory.Finally,it collected its trajectory points,with Q-learning theory to learn and find the optimal chasing path.In Visual Studio 2012 compiler environment,this paper got the credible motion animation of the agents and their emotional and physical factors＇ rules.Experimental results show that this method is efficient for promoting agents chasing.

关 键 词：情感计算 Q学习 博弈论 多智能体 自主追逐 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]