Walking Tracking System Based on Estimation of Human Posture in Omni-directional Mobile Walker

Abstract

In this paper, we develop a walking control system for a mobile body-weight support walker (i.e.Niltwamor;Novel Intelligent Lift-Type Walking-Assist Mobile Robot). Niltwamor is a walker capable of omni-directional movement and lifting control. First, the abdominal surface of the gait trainee is measured by the RGB-D camera, and the measured point cloud is subjected to elliptical fitting on the horizontal plane. Based on center position and direction of the approximated ellipse, the position coordinate and direction of the gait trainee are estimated. Based on that,Niltwamor tracks the trainee with proportional control. And therefore,it is possible to track the walking in all directions, and it is possible for Niltwamor to perform walking training with high degree of freedom without directly operating by the trainee. The effectiveness of the proposed system was verified by actual machine experiments for healthy subjects. With respect to estimation accuracy, the estimation error on the direction in the range of use is about 15[deg]. Then, as a result of tracking accuracy, the RMSE (Root Mean Square Error) of the position is about 9[cm] and the RMSE of the direction is about 8[deg].