一种对机器人阻抗控制中不确定性进行补偿的方法  

Method to Compensate Uncertainties in Robot Impedance Control

作　　者：王宇驰[1] 陈友东[1] 游玮 

机构地区：[1]北京航空航天大学机械工程及自动化学院,北京100191 [2]安徽埃夫特智能装备有限公司,安徽芜湖241007

出　　处：《机床与液压》2016年第9期7-9,37,共4页Machine Tool & Hydraulics

基　　金：国家863项目资助(2014AA041601)

摘　　要：机器人建模中存在的不确定性,给机器人精确控制带来了困难,在机器人力控制中尤为明显,制约了力控制在实际生产中的应用。采用模糊控制、神经网络等智能控制方法是解决这些经典控制理论所面临问题的有效手段。文中使用无监督学习的神经网络对不确定性进行在线补偿,提高阻抗控制的力跟踪性能,通过仿真验证了算法的有效性。The uncertainties existing in robot models present difficulties to controlling robots precisely. This is especially obvious in robot force control,and limits the usage of robot force control in industry field. Intelligent control,such as fuzzy control and neural network,is an effective method to solve this problem faced by classical control methods. Unsupervised learning network was adopted to compensate the uncertainties existing in robot impedance control online and to improve the performance of force tracking. The effectiveness of the proposed neural algorithm is verified by a simulation.

关 键 词：阻抗控制 神经网络 不确定性补偿 

分 类 号：TP242.2[自动化与计算机技术—检测技术与自动化装置]