移动机器人智能寻线导航与策略控制  

Intelligent line-tracking navigation and policy control of mobile robot

作　　者：张淑军[1] 孟庆春[1] 吴槟[1] 费云瑞[2] 

机构地区：[1]中国海洋大学计算机科学系,山东青岛266071 [2]中国海洋大学电子工程系,山东青岛266071

出　　处：《控制与决策》2005年第5期529-532,共4页Control and Decision

基　　金：山东省科技攻关课题(031080124).

摘　　要：寻线导航是移动机器人寻航方式之一,为使机器人沿地面标志线自主运动,并能在适时离线执行任务后自动返航,提出一种具有学习功能的智能寻线导航与策略控制方法以及系统模型,并将其应用于自行开发研制的导医机器人中.基于调制光检测原理,通过状态评判网络对多传感器信息进行融合,通过Q学习方法获得离线走失后的最优控制策略,实现了机器人在各种复杂未知环境下稳定、可靠地工作.实验结果表明,该方法可以有效地提高机器人的适应性和智能度.An intelligent line-tracking navigation and policy control method is proposed and the system model is designed in order to assist a mobile robot walking along an indicated line on the ground, to ensure it automatically navigate home after leaving the line and finishing its tasks. The method and model are applied to self-developed hospital guiding robot. Multi-sensory information is fused using a state-evaluation network based on modulated light detection principle. The optimal control strategy after straying from the line is achieved through Q-learning and the robot's steady and reliable work in different complex unknown environments is realized. The experimental results show that the proposed method can effectively enhance the robot's adaptability and intelligence.

关 键 词：自主移动机器人 寻线控制 BP网络 Q学习 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]