Feedback Iterative Learning Control for Rehabilitation Exoskeleton Robot Driven by Series Elastic Actuator^*

The escalating global prevalence of aging populations has positioned post-stroke limb dysfunction as a prominent determinant compromising quality of life among elderly populations, imposing substantial burdens on both patients and their families through multidimensional challenges encompassing physical disabilities, psychological distress, and socioeconomic pressures. As a novel intelligent rehabilitation device, the rehabilitation exoskeleton robot (RER) driven by series elastic actuator (SEA) can enhance patient engagement through precise power support and intelligent interaction mechanisms, promoting movement function recovery and enabling personalized therapeutic assistance. However, the safety, naturalness, and compliance control of the RER driven by SEA remain key challenges that need to be addressed. This paper proposes a reduced-order method for the dynamics model of the RER driven by SEA system based on singular perturbation theory to simplify its complexity. Additionally, a feedback iterative learning control (FILC) algorithm is designed, and its stability is theoretically analyzed. Numerical simulation results demonstrate that the proposed FILC algorithm exhibits excellent convergence speed, robustness, and stability in trajectory tracking tasks.