Neural network-based optimal parameter identification and variable-gain adaptive integral terminal sliding mode control for flexible joint robots

In this paper, a neural network (NN)-based adaptive integral terminal sliding mode controller (ITSMC) is proposed to control flexible joint robots (FJRs). To facilitate controller design, the high-order dynamics of FJRs are decomposed into two lower-order subsystems using the singular perturbation technique. System uncertainties are approximated by an NN, for which an optimal parameter identification (OPI) algorithm is developed to update the network weights, ensuring rapid convergence and high-precision uncertainty estimation. Based on these estimates, an ITSMC with adaptive feedback gains is designed to achieve high-precision trajectory tracking. The stability of both the identification and control schemes is rigorously analyzed and proven using Lyapunov stability theory. Simulation results validate the effectiveness and robustness of the proposed control strategy under model uncertainties and dynamic complexities.