Adaptive Control of Freedom Upper-Limb Exoskeleton Robots Using Extreme Learning Machine and Terminal Sliding Mode

This paper presents a novel Adaptive Dynamic Terminal Sliding Mode Controller (ADTSMC) with an exponential reaching law, specifically designed for precise position control of 5-Degree of Freedom (DOF) upper-limb exoskeleton robots. The proposed control strategy addresses critical challenges in precision, stability, and adaptability, introducing several key innovations. Firstly, the incorporation of a Terminal Sliding Mode Surface (TSMS) ensures finite-time convergence of system states, significantly enhancing tracking performance. Secondly, the integration of Dynamic Second-Order Sliding Mode Control (DSMC) with an exponential reaching law mitigates chattering, providing smoother and more reliable control inputs. Thirdly, an Extreme Learning Machine (ELM) is utilized to optimize controller parameters in real-time, offering robustness against uncertainties and variations in system dynamics while managing the complex nonlinearities of upper-limb exoskeletons. The ELM helps maintain stable learning and avoids issues such as gradient vanishing. The stability of the closed-loop system is rigorously validated using a Lyapunov candidate function, ensuring reliable operation under varying conditions. Simulation results show that the proposed adaptive DSMC outperforms existing methods in terms of convergence speed, tracking accuracy, and chattering suppression, demonstrating a significant improvement in performance, with reductions of approximately 74.3% in performance indices.