Estimation of seating face geometry during a curve using belt-shaped sensor system

Abstract

In this paper, we propose the estimation method of face geometry during a curve using a belt-shaped inertial sensor system in which multiple inertial sensors are mounted in a belt shape. In order to reduce the effects of dynamic acceleration and drift error during operation, the roll angle and pitch angle, which are the attitude information of the inertial sensor, are estimated using sensor fusion that focuses on velocity information including centrifugal acceleration during a curve motion. The shape of the entire belt-shaped inertial sensor system can be estimated by conducting forward kinematics analysis using the rotation matrix calculated from the attitude information and the distance between the sensors and applying the adaptive Kalman filter. The effectiveness of this method is evaluated by comparing the estimation result with the measurement result of the motion capture system. It is shown that high-precision shape estimation is possible by using this method even in an environment where dynamic acceleration acts significantly during curved motion.