Kinematic Modeling for Control of Agile Powered Prosthetic Legs Over Continuously Varying Speeds and Inclines

Abstract

People with above knee amputations face many unique challenges during their activities of daily living. Conventional passive prosthetic legs are not optimal for the range of ambulation tasks amputee users face daily, including walking at varied speeds and inclines, which hampers amputee mobility in the community and quality of life. Powered knee and ankle prosthetic legs have the potential to improve quality of by providing actuators that can perform net positive work at the knee and ankle, reducing the work required from the wearer and making more tasks possible. However, the controllers for these devices are limited to a small set of pre-defined tasks that require many hours of tuning for each user. The ubiquitous use of discrete task-specific controllers follows from the prevailing paradigm of viewing human locomotion as a discrete set of activities. The overall goal of this dissertation is to model human locomotion over continuously varying speeds and inclines to help realize the design of agile, powered prostheses without discrete task controllers. There is a fundamental gap in knowledge about how to analyze and model continuously varying locomotion, which greatly limits the adaptability and agility of powered prostheses. The central hypothesis of this dissertation is that the knee and ankle kinematics for a continuous interval of speeds and inclines can be parameterized by a continuous mathematical model based on gait phase,

walking speed, and incline alone. We have formulated a convex optimization framework to solve for the optimal parameters of this continuous model from a discrete experimental sampling of human kinematics during a variety of tasks. Using quasi-random phase shifting perturbations during a variety of walking inclines, we have investigated if a single phase variable can be used to accurately parameterize gait for a range of powered prosthetic leg applications. We then determined the degree to which sensors onboard the prosthesis can accurately measure walking speed and incline, and evaluate how the accuracy of these measurements affects the model’s ability to accurately predict joint kinematics for a number of users and conditions. This dissertation is scientifically significant to understanding how humans continuously adapt to speed and slope, technologically significant to the design of agile variable-activity prosthetic legs, and clinically significant to the adoption of powered prostheses that enable community ambulation for lower-limb amputees.