Estimation of Walking Direction from Wireless IMU Sensors for Path-Switching Control in Rail-Type Walking Support Robot

Abstract

Previously path switching was done manually by remote control on Yorisoi Robot, the walking support robot. In this paper, we have automated it by estimating the user’s walking direction. This is expected to enable users with walking disabilities to concentrate on their own walking. We estimate the user’s walking direction by using machine learning on the upper body movements obtained from the wireless IMU sensors. And, the training data (upper body movements and the user’s walking direction) are collected during the use of Yorisoi Robot in order to estimate the appropriate walking direction for each individual user. An experimental result showed that the method is able to successfully estimate the user’s direction about 93% and switch paths about 88%. This confirms that the estimation of the user ’s walking direction using wireless IMU sensors works well for path switching in Yorisoi Robot.