Adaptive Neural Network Control for the Cable-Driven Rolling Joint with Predefined Damping Ratio

Due to inherent compliance and safety, cable-driven robots have been widely used in the field of physical human-robot interaction. Scenarios that involve physical human-robot interaction impose stringent requirements on the transient performance of robots, which has become a key focus of current research. To tackle this problem, this paper designs a novel control strategy based the second-order Lyapunov stability criterion to predefine the damping ratio, thereby achieving an adjustable transient response. To alleviate the adverse impacts of unknown nonlinearities and external disturbances, neural networks trained via the gradient descent method are adopted in this paper for the online compensation of these nonlinearities. Furthermore, a new filter called dynamic command filter is utilized to resolve the "explosion of complexity"issue inherent in the traditional backstepping design approach, which arises from repeated differentiation of virtual signals. Based on the second-order Lyapunov stability criterion, this paper rigorously proves the practical stability of the closed-loop system and its ability to preserve the predefined damping ratio. Finally, the superiority of the proposed method is validated through comparative simulation experiments.