Adaptive Control of a Transtibial Prosthesis via Metaheuristic Optimization of a PD Controller

Significant advancements have been made in lower limb prostheses over the past decades. Nonetheless, there remain several areas where improvements can be achieved by simplifying the control architecture. This paper examines the implementation of a control architecture utilizing the Giza Pyramids Construction (GPC) optimization method to adaptively tune a basic Proportional-Derivative (PD) controller. As a fundamental aspect of any optimization problem, this work also emphasizes the introduction of an appropriate cost function, which significantly impacts the outcomes of the control strategy. The performance of the proposed control method is assessed using a simulated human model with a unilateral transtibial amputation. The simulation results show that the ankle joint can mimic the angle of a healthy subject with a normalized root mean square error (nRMSE) of 5.97 %. For five different model heights (ranging from 160 to 200 cm), this error remains below 7 % while keeping the hyperparameters of the optimization algorithm unchanged across all models. This relative insensitivity of the control strategy to human anatomy can be beneficial in rehabilitation clinics.