A study on the estimation method of 3D posture in body motion measurement using inertial sensors

Abstract

This paper proposes the measurement method of 3D posture using inertial sensors. The proposed method estimates the 3D posture (Roll-Pitch-Yaw angles in local coordinate) using the 3-axis angular velocity and the 3-axis acceleration. The 3D posture is compensated for the drift error of gyro sensor output by the translational and gravity accelerations of accelerometer output. The nonlinear state equation and the nonlinear measurement equation were established to estimate 3D posture. The Extended Kalman filter and the Unscented Kalman filter are applied to these equations. The measurement experiment was conducted to clarify the accuracy of proposed method using the experimental setup installing the rotary encoders. The results of the Unscented Kalman filter indicated higher than those by the Extended Kalman filter. In addition, the proposed method estimated the 3D posture compensating the drift error of gyro sensor in the measurement experiment generating the translational acceleration. Therefore, the proposed method can be used for the measurement of body motion. © 2013 The Japan Society of Mechanical Engineers.