Adaptive model-free control for ankle-assistive orthosis: A robust approach to real-time gait tracking
Oussama Bey, Yacine Amirat, Samer Mohammed
Abstract
Actuated Ankle-Foot Orthoses (AAFOs) assist dorsiflexion and plantarflexion movements at the ankle joint, supporting mobility and rehabilitation by complementing the wearer’s residual muscular activity within an assist-as-needed paradigm. Their effectiveness depends on advanced control strategies and accurate modeling of the coupled human-AAFO dynamics, which remains a challenging task. This paper presents a novel assist-as-needed control approach for an AAFO/wearer system based on an adaptive model-free framework, without the need for a dynamic model of the AAFO/wearer system. The proposed approach uses an ultra-local model, wherein a intelligent projection-based adaptive PID (iA-PID) controller is designed to achieve satisfactory tracking of a reference ankle joint trajectory. External torques affecting the AAFO/wearer system are estimated using a time-delay estimator and are compensated within the iPA-PID controller to ensure assist-as-needed control. Additionally, the projection operator constrains the evolution of the adaptive parameters, preventing actuator saturation and enabling controlled assistance delivery. Finite-time stability of the resulting closed-loop system is proven, and the final value theorem ensures that the tracking error converges to zero. The performance of the proposed approach is evaluated through simulations and real-time experiments with four healthy subjects. A comparison of tracking performance with several benchmark approaches was conducted as well as robustness tests under varying walking speeds to confirm the effectiveness and reliability of the proposed control approach.