Prescribed-Time Control of Flexible Joint Robotic Based on Neural Network

This paper proposes a neural network (NN) based adaptive prescribed-time sliding mode controller for flexible joint robotic (FJR) manipulators. By employing the singular perturbation technique, the original high-order system is decomposed into two lower-order subsystems. An NN is employed to estimate uncertainties within the FJR system, trained using a parameter identification algorithm (PIA). Subsequently, a prescribed-time sliding mode controller is developed for each subsystem to ensure that the tracking error converges within a predefined time and follows a given reference trajectory. The stability of the prescribed-time controller is rigorously established using Lyapunov analysis. Simulation and comparative experiments validate the effectiveness and superiority of the proposed method.