Neural Adaptive Control for Robotic Systems With Saturation and Disturbance

This paper proposes a neural adaptive control (NAC) strategy for the robotic systems with saturation and disturbance. The strategy involves using a neural network (NN) to estimate the robot’s dynamic model and constructing a compensator to mitigate the impact of input saturation. To obtain estimated velocities and external disturbances in the robotic systems without prior knowledge of the system model, an NN-based extended state observer (NNBESO) is designed. Based on the designed error compensation term and NNBESO, NAC is proposed to achieve the tracking for the desired trajectory under input saturation and external disturbance. Finally, the proposed NNBESO-based NAC strategy is verified by the simulation experiments on a 2 rigid-link robotic manipulator with input saturation and system disturbances, and the results demonstrate the effectiveness of the proposed NNBESO-based NAC strategy.