人工情感的进化控制系统实现  

Implementation of an evolutionary control system based on artificial emotion

作　　者：徐雄[1] 

机构地区：[1]湖北师范学院,湖北黄石435002

出　　处：《智能系统学报》2008年第2期135-139,共5页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金资助项目(50275150);教育部博士学科点专项科研基金资助项目(20040533035);湖北师范学院研究生科研启动基金

摘　　要：人工情感在机器人的研究中至关重要,简要概括了当前人工情感的应用.在借鉴情感学习控制的理论的基础上,融入了进化控制的思想,设计出了一种基于人工情感的控制体系结构,在此结构中包含有基于遗传算法的进化控制系统、神经和人工情感控制系统.机器人通过神经系统接受环境信息并进行行为决策,行为决策的效果通过情感学习模型进行反馈.情感学习模型根据机器人的内、外环境状态,产生情感因子(即生物激素),再由情感因子来调节神经系统的记忆和行为决策,最后神经系统的记忆与行为模块又由进化系统得以继承.该控制结构加强了机器人在动态环境中的学习和自适应能力.仿真实验验证了该控制结构的有效性,仿真结果也表明机器人具有很强的学习和自适应能力.Artificial emotion is essential to robotics study. Typical implementation of artificial emotion is briefly summarized in this paper. In order to better implement emotion-learning control mechanisms, a new control architecture based on artificial emotion is proposed. It incorporates an evolution-control sys- tem based on a Genetic Algorithm with a neural and an artificial emotion control system. The neural system receives environmental information and makes decisions. The results of decision-making are fed back to the emotion-learning model. The emotion-learning model produces emotional factors （hormones） based on inner and outer conditions of the robot, and these factors is used to regulate the neural system. In the final step, the memorization and behavior module of the neural system is exported to the genetic environment. This control architecture enhances learning and adaptive capacities of robots in a dynamic environment. Simulation was made to confirm the validity of the control architecture.

关 键 词：神经网络 遗传算法 人工情感 行为决策 机器人 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]