Measurement of apple harvesting behavior applying estimation error reduction method from body-mounted inertial sensor by low pass filter

Abstract

In this study, a measurement method of apple harvesting behavior to design support suit which helps harvesting work is proposed. In this method, ten inertial sensors which can measure triaxial acceleration, triaxial angular velocity, and triaxial geomagnetic are mounted on farmer's chest, waist, right and left upper arms, forearms, thighs, and shanks. Using these inertial sensors, two estimation methods are applied to know three-dimensional direction of body link; one is to use acceleration and geomagnetic, and the other is to integrate angular velocity. Estimation value derived from acceleration and geomagnetic includes high frequency noise which originate from body motion and magnetic disturbances. Estimation value obtained from integration of angular velocity includes low frequency noise which originate from sensor drift. To reduce estimation error of the direction of body link resulting from integration of angular velocity, acceleration and geomagnetic is applied. Discontinuity of posture of body links after the error correction is solved by a low pass filter. Harvesting work is measured and estimation results indicate that the arm movement differs according to the type of work.