Understanding Individuality in Human Locomotion for Configuring Powered Prosthetic Legs

The development of powered prosthetic legs has the potential to improve the mobility and quality of life for the population of people with amputation. The proliferation of research in this field promises that we will have not only agile robotic legs in the near future, but also devices that can seamlessly integrate into our daily life–to smoothly react to all the different terrains and activities that we encounter every day. Although the field has shown great progress on all fronts since the original research devices, there are still a number of hurdles before we reach devices like those we see in science-fiction. This research centers around improving the clinical viability of powered prostheses, specifically the biomechanical grounding and control methodology of clinically implementing a powered prosthetic leg for an individual. This research breaks that problem down into four main sections: dataset development, individual trend identification, individualized modeling, and experimental validation. First, we present the development of a multi-modal, able-bodied dataset designed to comprehensively present the movement patterns of multiple activities of daily living and the potential nuances within each activity. Then, the dataset is investigated to identify trends in how individuals tackle a wide range of tasks. Next, leveraging these trends, we can quickly and efficiently model the kinematics and kinetics of an able-bodied person with an eye towards designing individualized kinematic and impedance controllers for powered prosthetic legs. Lastly, we discuss the development and experimental validation of a clinical tuning interface that allows a prosthetist to tune complex, multi-terrain controllers without the assistance of an engineer. This work is the first to clinically interface with continuous hybrid kinematic impedance control, successfully individualizing a powered prosthesis in a clinically relevant and viable way.PHDRoboticsUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/177965/1/reznick_1.pd