Active and Adaptive Human-Robot Interaction Control for Robot-assisted Gait Training

An active and adaptive human-robot interaction control method is proposed, in which active gait trajectory planning, stiffness adaptive control, and gait replanning are integrated to enable compliant interaction and ensure patient safety during robot-assisted gait training. Firstly, a parameterized model of gait trajectory is constructed using finite Fourier series, to address the problem of insufficient patient engagement in gait training. An online gait trajectory planning method is developed based on surface electromyography (sEMG) signals, by which intention-driven active gait trajectory planning is achieved. Secondly, a stiffness adaptive control strategy is proposed based on position deviation and muscle activation. A fuzzy control algorithm is employed to adjust the equivalent joint stiffness of the robot in real time, whereby the compliance of human-robot interaction can be effectively improved. Moreover, evaluation indices for dynamic balance of the human-robot system and trajectory stability are established, to mitigate the risks of motion imbalance and potential falls during clinical applications. Based on the online evaluation results, gait trajectory re-planning is carried out using model predictive control. Finally, the effectiveness of the proposed method is validated through system simulations and practical experiments.