Gait Phase Recognition Based on A Wearable Depth Camera

Abstract

Gait phase recognition is fundamental for the control of assistive lower-limb exoskeletons or prostheses. In this paper, we have proposed an innovative strategy to estimate the human walking gait phase by means of a wearable depth camera. The proposed system is composed by two subsystems: periodic depth signal extraction and adaptive oscillator-based gait phase estimation. Validation experiments have been implemented with four subjects. Each subject performed three free ground-level walking trials at his/her preferred speed. Results showed that the proposed system could provide an accurate gait phase estimation based on a stable and periodic gait-related depth signal. The promising performance is expected to enable a lower-limb wearable robot to provide more stable and effective assistance for daily walking tasks.