In this chapter, we consider the obstacle avoidance problem of redundant robot manipulators with physical constraints compliance, where static and dynamic obstacles are investigated. Both the robot and obstacles are abstracted as two critical point sets, respectively, relying on the general class-K functions, the obstacle avoidance problem is formulated into an inequality in speed level. The minimal-velocity-norm (MVN) is regarded as the cost function, converting the kinematic control problem of redundant manipulators considering obstacle avoidance into a constraint-quadratic-programming problem, in which the joint angles and joint velocity constraints are built in velocity level in form of inequality. To solve it, a novel deep recurrent neural network based controller is proposed. Theoretical analyses and the corresponding simulative experiments are given successively, showing that the proposed neural controller does not only avoid collision with obstacles, but also track the desired trajectory correctly.


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Deep RNN Based Obstacle Avoidance Control for Redundant Manipulators


    Beteiligte:
    Zhou, Xuefeng (Autor:in) / Xu, Zhihao (Autor:in) / Li, Shuai (Autor:in) / Wu, Hongmin (Autor:in) / Cheng, Taobo (Autor:in) / Lv, Xiaojing (Autor:in)

    Erschienen in:

    Erscheinungsdatum :

    03.06.2020


    Format / Umfang :

    19 pages




    Medientyp :

    Aufsatz/Kapitel (Buch)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch





    An obstacle avoidance scheme for planar hyper-redundant manipulators

    Ma, S. / Konno, M. | British Library Online Contents | 1997


    Obstacle avoidance for redundant robots using configuration control

    Seraji, Homayoun / Colbaugh, Richard D. / Glass, Kristin L. | NTRS | 1992



    Collision Avoidance of Cooperative Dual Redundant Manipulators

    ZHAN, Qiang / HE, Yan-hui / CHEN, Ming | Elsevier | 2003

    Freier Zugriff

    Towards Dynamic Obstacle Avoidance for Robot Manipulators with Deep Reinforcement Learning

    Zindler, Friedemann / Lucchi, Matteo / Wohlhart, Lucas et al. | Springer Verlag | 2022