Enhanced Orientation Tracking for Redundant Manipulators via DNN-Based Double Control

This paper proposes a new double-indicator control scheme of redundant manipulators, which utilizes an untrained dynamic neural network (DNN) solver. The control scheme combines direction tracking, physical constraints, and anti-noise design to address problems of the high computational complexity and the lack of direction tracking in existing neural network-based solutions. In addition, the DNN solver provides a controltheoretic framework which ensures the global and exponential convergence, stability, and robustness of the control scheme. In our design, we specifically consider the effect of noises on the system and incorporate the anti-noise mechanism. Furthermore, the effectiveness and feasibility of the proposed control scheme are verified through simulations with a KUKA LBR iiwa 7 R800 manipulator. The results show that the DNN-based doubleindicator control scheme efficiently generate accurate motion trajectories while maintaining the directional stability of the endeffector, and resist noises.