基于内分泌调节机制的机器人行为规划算法及其应用研究  

Motion-planning Algorithm Based on Endocrine Regulation Mechanism and its Application on RoboCup

作　　者：黄国锐[1] 徐敏[1] 张荣[1] 曹先彬[1] 王煦法[1] 

机构地区：[1]中国科学技术大学计算机科学技术系,安徽合肥230026

出　　处：《小型微型计算机系统》2004年第2期262-265,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目 (699710 2 2 )资助;博士点基金资助;博士点基金 (19990 3 5 82 6)资助

摘　　要：借鉴内分泌系统对神经系统与遗传系统的高层调节机制 ,提出了一种新的基于内分泌调节机制的机器人行为规划算法 .此算法中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一种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式又由遗传系统得以继承 .该算法有效避免了神经系统复杂的自学习过程 ,同时也保证机器人有较强的自适应能力 .为了验证算法的有效性 ,本文做了机器人足球队守门员训练的仿真实验 ,结果也表明该算法具有很强的自适应学习能力 .Self-adaptation of robot in dynamic and complex environment is a hot topic in the field of intelligent control. Motivated by the high-level regulation of endocrine system to neural system and genetic system, this paper puts forward a new motion-planning algorithm based on endocrine regulation mechanism. In this algorithm, neural system receives environmental information and makes decision. The result of decision-making is fed back by emotion-learning model. The emotion-learning model produces emotional factors (hormones) based on inner and outer conditions of the robot, then uses these factors to regulate neural system, at last the memorization and behavior mode of neural system is exported to genetic environment. So the algorithm avoids the complex self-study of neural system and gets better performance. In order to prove the validity of this algorithm, we use it to control an inverted pendulum. The result also shows strong self-adaptation performance of the algorithm.

关 键 词：内分泌系统 行为规划 情感学习 神经网络 遗传算法 足球机器人 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]