Study on the neural network impedance force control for inner-wall grinding robot

摘要：

On the basis of adaptive learning capability of NN from data directly,this paper presents an accurate and computationally efficient scheme for force tracking on uncertain contact environment using multilayer perceptron NN-based position impedance control system.Through formula deduction,the NN sample inputs are set as the actual contact trajectory points and contact forces,and the NN sample outputs are set as the environment predictive trajectory points,stiffness and tangential angle to train NN.Then according to the desired force and force-error,the NN-based position impedance control system can estimate the required reference position on-line.Moreover,a comprehensive simulation study has been carried out using the solid propellant missile wall grinding robot,and the simulation studies demonstrate that the scheme is able to compensate for uncertainties of both the environmental trajectory and stiffness,and the end-effector can exert the desired contact force to the uncertain environment with the real-time compensation of impedance control.

关键词： force tracking NN-based impedance control solid propellant missile wall grinding robot

作者: Yi-lan Sun Hong-yi Liu Dan-dan Cui

作者单位: Department of Mechanical Engineering & Automation,University of Northeastern,Shenyang,Liaoning Province,China

会议类型: 国际会议

会议名称: International Conference on Modelling,Identification and Control(模拟、鉴定、控制国际会议)

会议地点: 上海

会议语种:英文

在线出版日期: 2008-06-29（万方平台首次上网日期，不代表论文的发表时间）