IR vision system for the estimation of gait phase of the MIT-Skywalker

Abstract

The MIT-Skywalker is a robotic device intended to promote gait rehabilitation. It makes use of the concept of passive walkers, using gravity to actuate the leg during the swing phase. The timing and profile of the robot actuation relies on gait phases triggered by the subject's leg position. We employ a vision system to determine these phases. In this paper, we present this novel vision system employed to determine the two-dimensional gait kinematics in the sagittal plane. The system uses a single high resolution camera running at a high frame rate to enable human-in-the-loop control of the robot. Operating solely in the infrared spectrum, the vision system is able to be used in cluttered clinical environments. Here we characterize the performance of this vision system, showing angle errors of less Than 1° with a loop time smaller than 10 msec Such accuracy and speed allow smooth motion control of the MIT-Skywalker and can also log high resolution gait data.