Orientation Tracking Incorporated Multicriteria Control for Redundant Manipulators With Dynamic Neural Network

Existing neural-network-based solutions for controlling a redundant robot are trapped by the relatively high computational complexity and the lack of the incorporation of orientation tracking. In order to remedy these two weaknesses, this article proposes a new multicriteria control scheme aided with a training-free dynamic neural network (DNN), which simultaneously considers the orientation-tracking constraint and physical constraints. Meanwhile, compared with existing methods for handling the same task, the proposed DNN solver is of low computational complexity. Theoretical analyses confirm that the proposed scheme based on the DNN solver globally and exponentially converges to the theoretical solution of the robotic motion generation. Besides, illustrative simulations and physical experiments based on a Franka Emika Panda manipulator demonstrate the validity and feasibility of the proposed scheme with the DNN solver.