Enhancing Control of Lower Limb Prostheses by Decoding Intention and Detecting Gait Perturbations

This thesis focuses on developing novel high-level control algorithms for lower limb prostheses to enhance their usability, reliability, and efficacy. Permitted movements, reliability, and overall usability are crucial factors in prosthesis acceptance as they directly influence the user’s independence. Focusing on these essential aspects can improve embodiment, foster agency, and ultimately lower the likelihood of early abandonment. Therefore, this thesis pays particular attention to locomotion mode classification, intention detection, and perturbation detection algorithms. After reviewing the existing technologies in the field of lower limb prostheses and trip detection, I present different intention decoding algorithms for locomotion mode classification, gait initiation, and step direction prediction. For each approach, I discuss the specific application, real-time implementation, and potential limitations for clinical adoption. I also present our approach to studying and detecting trip perturbations for users of lower limb prostheses, presenting the validation of a novel experimental setup for eliciting trip perturbations in a controlled environment and different trip detection approaches.