Walking is a metabolically demanding activity. Scientists have developed two general approaches to reducing the metabolic cost of walking using exoskeletons:​ 1) adding energy to the human-device system to assist concentric muscle contractions, and 2) transferring energy from one gait phase to another (or from one joint to another) to assist isometric muscle contractions. Despite several exoskeletons' success in reducing the metabolic cost of walking, the practical application of these devices is limited by their short operational lifespans and poor adaptive control.

This thesis outlines the design and evaluation of a novel lower limb exoskeleton capable of reducing the metabolic cost of walking. This device applies a novel approach to user assistance—removing energy to assist the hamstrings’ eccentric contractions during the swing period. Although devices that assist with concentric and isometric contractions have yielded larger metabolic cost reductions, we show that assisting with eccentric contractions can significantly reduce metabolic cost with the concurrent benefit of electricity production. This capacity to convert removed energy into electricity suggests that devices employing this assistive technique can be self-powering, extending their operational lifespan and enabling adaptive control. We also provide direct evidence that the assistance profile's timing and magnitude are important factors in reducing the metabolic cost of walking. We demonstrate that applying a moderate level of assistance during the terminal swing phase results in the greatest metabolic cost reduction under the present experimental conditions.

The present work advances our understanding of the complex human-device interactions that occur during walking with an exoskeleton, as well as the general principles governing human gait. In addition to elucidating the energetics and biomechanics of walking, the present work provides insights into the design, control, and evaluation of devices that remove energy for their users during walking. Based on our findings, we recommend that future lower-limb exoskeletons employ muscle-specific parameters and profiles to provide the largest amount of metabolic assistance.