Estimation method of operating force using an extended Kalman filter based on ground reaction force and human behavior

Abstract

Aiming to assist operators, our laboratory had proposed power assisted wire suspension system. A system to estimate the operating force based on ground reaction force and human behavior has been developed. However, it is necessary to measure the human behavior (position, velocity, and acceleration) of 17 points on the body with the conventional human dynamics model. Therefore, the system is extremely expensive and not practical, because of large number of sensors. In addition, fewer sensor units worsen the estimation accuracy of the operating force. In order to improve the estimation accuracy of the operating force, it is important to identify positions of center-of-gravity and masses of links. Thus, this paper presents the novel human dynamics model estimating the operating force and parameters using extended Kalman filter based on human behavior with only 6 sensor units and ground reaction force. Improvement of the estimation accuracy is confirmed by experiments.