基于增强式学习的微小型移动机器人位姿控制  

Control of a Micro-Mobile Robot for Position and Orientation Problem Based on Reinforcement Learning

作　　者：李小海[1] 肖永利[1] 程君实[1] 

机构地区：[1]上海交通大学信息存储研究中心,上海200030

出　　处：《上海交通大学学报》2000年第11期1478-1482,共5页Journal of Shanghai Jiaotong University

基　　金：国家自然科学基金重点资助项目! ( 698890 50 )

摘　　要：根据自行设计的两轮独立驱动的微小型移动机器人的特点 ,利用增强式的学习方法 ,对输入量进行分层模糊化以确定模糊区域 ,然后选取相应区域中具有重大适应值的动作控制机器人移动 ,得到了该机器人在无约束空间中时间优化的移动路径 ,实现了该移动路径的自动获取 。According to the characteristic of the micro mobile robot with two independent wheels, this paper proposed a control method for its position and orientation problem. This method is based on reinforcement learning. It uses the hierarchical fuzziness of the inputs to get the fuzzy area. Every fuzzy area has three kinds of actions. The micro mobile robot moves according to the action with the maximal performance in this fuzzy area. This method gets the moving route automatically within the optimal time in unrestraint space. The imitation and experimental results show this method is feasible.

关 键 词：移动机器人 位姿控制 增强式产 分层模糊化 

分 类 号：TP242.602[自动化与计算机技术—检测技术与自动化装置]