NRI: Balance Recovery Control for Amputees Using Powered Leg Prostheses

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

Falls among amputees are frequent, costly and negatively impact quality of life. In 2005, there were one million people in the United States with lower limb amputation. About 10 percent of these people experience at least one fall that results in serious injury, resulting in estimated costs to the United States health care system of about $1.1 billion annually. This number is expected to quadruple by the year 2050 due to increasing rates of obesity and diabetes. People with amputation are conscious of their increased fall risk, leading to reduced mobility and social activity. About half of the amputee population reports a fear of falling. Similarly large numbers list as major limitations to their quality of life the inability to walk on uneven terrain or without a stabilizing gait aid. These facts highlight the challenge imposed by imbalance, and suggest that improving balance recovery in amputee locomotion would significantly improve the quality of life of lower-limb amputees as well as reduce related health care costs. The emerging field of robotic prosthetics provides the opportunity to attack this problem with novel prosthesis designs and control strategies. The project seeks to take advantage of this opportunity. It combines computational models of the human neuromuscular control system with novel designs of powered knee-and-ankle prostheses and biomechanical gait analysis to establish new control paradigms for powered prostheses that substantially improve the ability of amputees to recover from large disturbances. Implemented in commercial devices, these control techniques could reduce fall rates and fear of falling, thereby improving the mobility and quality of life for millions of people. Other benefits of the project include the support of education. Graduate and undergraduate students will be trained and mentored, and research tools and outcomes will be integrated into coursework, including the software and hardware tools developed in this project for studying prosthesis control and disturbance recovery in human locomotion. In addition, the project supports the dissemination of research results by publishing and freely providing simulation code, control implementation code, and hardware designs online. The overarching goal of this project is to test the hypothesis that a reflex-like prosthesis control strategy inspired by human motor control substantially improves the balance recovery for above-knee amputees during walking. Balance recovery has evolved into a major research area as fall-connected injuries are one of the main causes of impairment, disability and death in aging societies. Lower limb amputees are especially at risk of falling as current prosthetic limbs provide only limited functionality for recovering from unexpected disturbances. The project combines methods from computational neuromechanics, robotic prosthetics, and biomechanical gait analysis to identify prosthesis control strategies that help above-knee amputees recover balance after large disturbances such as trips, slips and pushes. An existing reflex control model of human locomotion is adapted to amputee gait, involving theoretical research on feedback control algorithms for powered prosthetic limbs and predictions of amputee recovery behavior in simulated experiments. Prototypes of powered knee-ankle prostheses are developed, including a tethered prosthesis emulator for rapid human-in-the-loop control design and evaluation on a treadmill, and a mobile prosthesis allowing evaluation outside the laboratory. Control algorithms identified in the reflex control model are embedded in these prototypes and systematically evaluated in balance recovery experiments with above-knee amputees. An outcome confirming the hypothesis could establish new control paradigms for powered prostheses and enable practical controllers for improved balance recovery in amputee gait. In addition, the project will advance theoretical models of human balance recovery as well as control algorithms and hardware designs for robotic knee-ankle prostheses.