Personalized Gait Generation and Active Compliance Control Strategy for Lower Limb Rehabilitation Robots

Home-based lower limb rehabilitation robots (HLLRRs) serve as a valuable complement to clinical therapy by enabling rehabilitation in a home setting. However, ensuring these robots provide safe, personalized, and functional level-appropriate training remains a significant challenge. To address this issue, this study proposes a personalized gait generation method to deliver tailored trajectories and introduces a two-loop active compliance control strategy to integrate patients’ motion intentions into the robot’s control for robot-assisted training. Specifically, individual differences in physical body parameters are mapped into gait pattern variations using a convolutional neural network (CNN)-based gait generation model trained with RepOptimizer. The outer-loop trajectory planner incorporates the wearer’s motion intention into the change of step time according to the interaction torque and predicted system states. The inner-loop position controller is designed based on the active disturbance rejection control (ADRC) strategy. Simulation results demonstrate that the generated trajectories closely align with normal gait patterns, with loss within 0.13017, making them suitable as reference trajectories for robot-assisted walking training. The replanned trajectories remain smooth and preserve the characteristics of the original ones under various interaction torques. The designed controller tracks the desired trajectory quickly and steadily, with a mean absolute error (MAE) of less than 0.03107 rad, ensuring reliable performance throughout the training.