Development and validation of a terrain adaptive prosthesis control system

Abstract

Wearable lower limb robotic devices have great potential in addressing gait pathologies through assistive or rehabilitative means. In the case of amputation, powered prostheses can be used to recapitulate biological walking, improving mobility and diminishing amputation-associated comorbidity. In the case of intact limb pathologies such as weakness or paralysis, powered exoskeletons can be used for similar goals. A major challenge in developing these technologies lies in their control, whose aim is to improve gait dynamics across a variety of walking conditions. Perhaps the most significant determinant of gait dynamics is ground terrain: numerous studies have shown that walking on level ground, inclines, or stairs significantly affects leg dynamics. Additionally, it has been shown that abnormal or asymmetrical gait across any of these conditions causes comorbidities secondary to gait pathology, including back pain, increased fatigue, and in the case of amputation, osteoarthritis of joints in the unaffected limb. Motivated by the desire to normalize gait mechanics across a variety of conditions, the principle aim of this work is to develop an automatically terrain adaptive control system for lower limb robotic devices, wherein the control system anticipates transitions in walking tasks independently of external devices and switches to corresponding control policies. In particular, we focus on development and validation of such a control system in a below-knee prosthesis. The final result of this work is a method to automatically measure and accurately predict terrain geometry in a lower limb robotic device as a person is walking, along with a terrain-adaptive tunable control model that can successfully improve gait dynamics across multiple walking conditions.