Adaptive Compliance Control for a Wearable Lower Limb Rehabilitation Robot Based on Online Estimation of User's Joint Stiffness

Although numerous variable stiffness controllers have been proposed, generating compliant and adaptive trajectory that could synchronize with human multiple joints movement remains an open challenge for wearable lower limb rehabilitation robots. In this study, we proposed a new adaptive compliance controller (ACC) for a bilateral hip and knee wearable robotic system via online estimation of users' joints stiffness along with adaptive trajectory deformation algorithm (TDA). A sEMG-driven biarticular Hill-type neuromuscular model was first developed and calibrated to estimate the hip and knee joints’ stiffness. The adaptive TDA was designed to transfer joint stiffness adjustments from human joints for robots to reshape the trajectories for multiple joints smoothly. Robot-assisted overground walking experiments were conducted to comprehensively compare our proposed method with similar approaches, including force-based adaptive compliance, fixed-parameter compliance, and passive control, among five healthy human subjects. The results demonstrated that the proposed ACC outperformed other controllers in matching the desired movements of multiple human joints, showing superior performance in tracking error, muscle activity, and gait symmetry and coordination. These findings emphasized that ACC plays a key role in enabling human-robot synchronization across multiple joints, offering significant potential to improve the efficiency of robot-assisted gait rehabilitation.