Most commercially available prosthetic feet do not exhibit a biomimetic torque-angle relationship, and are unable to modulate their mechanics to assist with other mobility tasks, such as stairs and ramps. In this paper, we present a quasi-passive ankle-foot prosthesis with a customizable torque-angle curve and an ability to quickly modulate ankle stiffness between tasks. The customizable torque-angle curve is obtained with a cam-based transmission and a fiberglass leaf spring. To achieve variable stiffness, the leaf spring's support conditions can be actively modulated by a small motor, shifting the torque-angle curve to be more or less stiff. We introduce the design, characterize the available torque-angle curves, and present kinematics from a transtibial amputee subject performing level-ground walking, stair ascent/descent, and ramp ascent/descent. The subject exhibited a more normative range of motion on stairs and ramps at lower stiffness levels, and preferred different stiffness levels for each task. Paired with an appropriate intent recognition system, our novel ankle prosthesis could improve gait biomechanics during walking and many other mobility tasks.

BackgroundProsthetic feet are spring-like, and their stiffness critically affects the wearer’s stability, comfort, and energetic cost of walking. Despite the importance of stiffness in ambulation, the prescription process often entails testing a limited number of prostheses, which may result in patients receiving a foot with suboptimal mechanics. To understand the resolution with which prostheses should be individually optimized, we sought to characterize below-knee prosthesis users’ psychophysical sensitivity to prosthesis stiffness.MethodsWe used a novel variable-stiffness ankle prosthesis to measure the repeatability of user-selected preferred stiffness, and implemented a psychophysical experiment to characterize the just noticeable difference of stiffness during locomotion.ResultsAll eight subjects with below-knee amputation exhibited high repeatability in selecting their Preferred Stiffness (mean coefficient of variation: 14.2 ± 1.7%) and were able to correctly identify a 7.7 ± 1.3% change in ankle stiffness (with 75% accuracy).ConclusionsThis high sensitivity suggests prosthetic foot stiffness should be tuned with a high degree of precision on an individual basis. These results also highlight the need for a pairing of new robotic prescription tools and mechanical characterizations of prosthetic feet.

No abstract