Adaptive Passive Control for a Lower Limb Rehabilitation Robot Based on Controller Gain Tuning

Background: The inherent compliance of cable-driven lower limb rehabilitation robots, though advantageous for safe human-robot interaction, can lead to unsatisfactory trajectory tracking performance when control strategies are inadequately designed, hindering their effectiveness in clinical rehabilitation. Methods: This study presents an adaptive passive control strategy based on online adjustment of controller gains. A dynamics model of the robot was established using Lagrange’s equations. A gain-adaptive iterative learning control algorithm was proposed, and its stability was rigorously proven via Lyapunov analysis. Reference trajectories for the hip and knee joints were generated from captured human gait data. The method was validated through extensive simulations and experiments on a self-developed rehabilitation robot prototype. Results: Simulation results demonstrated that the proposed method achieves convergence nearly five times faster than conventional iterative algorithms under various disturbances, with mean tracking errors of 1.24% for the hip and 0.321% for the knee. Experimental results on the physical system showed an average root mean square error of 0.166 cm, a mean absolute error of 0.134 cm, a maximum error of 0.396 cm, and a tracking percentage error of 5.62%. Conclusions: The proposed control method significantly improves trajectory tracking accuracy in cable-driven lower limb rehabilitation robots, showing strong potential for application in clinical rehabilitation settings. The method enhances tracking precision and convergence speed, indicating its value for supporting recovery in patients with motor impairments.