Mobile Robot Assisted Gait Monitoring and Dynamic Margin of Stability Estimation

Abstract

To assess balance control and fall risk, it is desirable to continuously monitor dynamic stability during walking tasks. Dynamic Margin of Stability (MoS) is widely recognized as a quantitative measure for human walking stability and gait balance strategies. We propose a mobile robot assisted gait monitoring system that precedes human subjects in overground walking. Real-time data from the RGB-D Kinect sensor on the robot are fused with measurement from pressure sensors and inertial measurement units in a pair of instrumented footwear, and Kalman filter based methods are developed to estimate MoS and spatiotemporal gait parameters in real time. Experimental results with 10 subjects are compared with those obtained by a gold-standard motion capture system. Results show that the proposed method achieves acceptable accuracy of MoS estimation and high accuracy for spatio-temporal gait parameters. Whereas existing works on MoS assessment use wearable sensors that can only provide offline analysis, our proposed system provides real time gait monitoring and MoS estimation that could potentially assess fall risk during walking in out-of-lab conditions.