Fusion Control of a Lower Limb Hybrid Neural Prosthesis Based on Model Predictive Control and Sliding Mode Control Strategies

Functional electrical stimulation (FES) and powered exoskeletons are widely applied methods for lower limb rehabilitation. These approaches assist human movement by utilizing the body's own muscles along with motors in the exoskeleton. However, the complexity and uncertainty of system modeling hinder the widespread adoption of model-based control schemes. This paper proposes a hybrid control scheme to track the dynamic knee joint angle and provides a Lyapunov stability proof. The scheme initially optimizes the control input using model predictive control (MPC), supplemented by sliding mode control (SMC) to provide compensatory torque against system disturbances. Simulation results show that, compared to PID control, the new method achieved a 55.1% reduction in the sum of squared torques (SST) at the cost of an increase of 0.55° in RMSE. Compared to NMPC control, the new method reduced RMSE by 41.2% while increasing SST by 2.8%. The fusion control method retains the advantages of the original NMPC while achieving a significant reduction in tracking error with a relatively small increase in motor torque, demonstrating higher robustness.