Among tracking control of redundant manipulators, potential limitations such as model uncertainties and physical limitations may exist. Conventional solutions may fail when model parameters differ from nominal ones. In this chapter, a novel kinematic controller with the capability of self-adaptation is proposed to address this challenging issue. Based on the coordinate feedback, a Jacobian adaption strategy is firstly built by updating kinematic parameters online. Using Karush–Kuhn–Tucker conditions, the redundancy solution problem is then turned into a quadratic optimization one, and a recurrent neural network based controller is designed to derive the optimal solution recurrently. Theoretical analysis demonstrates the global convergence of the tracking error. Compared with existing methods, kinematic model uncertainty of the robot is allowed in this chapter, and the pseudo-inverse of Jacobian matrix is avoided, with the consideration of physical limitation in a joint framework. Numerical experiments based on Kinova JACO 2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$_2$$\end{document} show the effectiveness of the proposed controller.


    Access

    Download


    Export, share and cite



    Title :

    RNN Based Trajectory Control for Manipulators with Uncertain Kinematic Parameters


    Contributors:
    Zhou, Xuefeng (author) / Xu, Zhihao (author) / Li, Shuai (author) / Wu, Hongmin (author) / Cheng, Taobo (author) / Lv, Xiaojing (author)

    Published in:

    Publication date :

    2020-06-03


    Size :

    22 pages




    Type of media :

    Article/Chapter (Book)


    Type of material :

    Electronic Resource


    Language :

    English