Different-layer control of robotic manipulators based on a novel direct-discretization RNN algorithm

• Published in: Neurocomputing (Amsterdam) Vol. 620; p. 129252
• Main Authors: Guo, Jinjin, Xiao, Zhanhao, Guo, Jianhua, Hu, Xianglei, Qiu, Binbin
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.03.2025
• Subjects: Convergence performance; Direct-discretization technique; Discrete time-dependent different-layer control; Recurrent neural network; Robotic manipulators; Convergence performance; Recurrent neural network; Robotic manipulators; Direct-discretization technique; Discrete time-dependent different-layer control
• ISSN: 0925-2312
• DOI: 10.1016/j.neucom.2024.129252

In this paper, the development of discrete-time recurrent neural network (RNN) algorithm is different from previous studies that require the derivation processes in the continuous-time environment. Specifically, the problem of discrete time-dependent different-layer control (DTDDLC) of robotic manipulators is investigated. To solve the DTDDLC problem, a discrete-time RNN algorithm based on direct-discretization technique is proposed, and thus named direct-discretization RNN algorithm for convenience. For comparative purposes, four indirect-discretization RNN algorithms are also developed to solve the same problem. Theoretical analyses and results manifest the convergence performance of the proposed direct-discretization RNN algorithm. Finally, numerical experiments based on a five-link planar robotic manipulator and a UR3 robotic manipulator substantiate the effectiveness and superiority of the proposed direct-discretization RNN algorithm.