I. INTRODUCTION
Over 10 million people in the United States currently suffer from significant long-term disability due to stroke, or are living with a lower limb amputation [1], [2]. This population could benefit from robot-assisted rehabilitation and wearable devices, as a mean to regain functional mobility. Since gait support systems need to be synchronized with the human motion [3],[4], precise characterization of the walking phases is critical. Human gait is separated into gait cycles, defined as the intervals between consecutive heel-strikes of the same foot [5]. To ensure proper functionality of the gait support systems, accurate and robust gait segmentation is necessary.