Real-time implementation of orientation correction algorithm for 3D hand motion tracking interface

Abstract

This paper outlines the real-time implementation of an orientation correction algorithm using the gravity vector and the magnetic North vector for a miniature, commercial-grade Inertial Measurement Unit to improve orientation tracking in 3D hand motion tracking interface. The algorithm uses the sensor fusion approach to determine the correct orientation of the human hand motion in 3D environment. The bias offset error is the IMU’s systematic error that can cause a problem in orientation tracking called drift. The algorithm is able to determine the bias offset error and update the gyroscope reading to obtain unbiased angular velocity. Furthermore, the algorithm will compare the initial estimated orientation result by using other referencing sources which are the gravity vector measured from the accelerometer and the magnetic North vector measured from the magnetometer, resulting in the improvement of the estimated orientation. The orientation correction algorithm is implemented in real-time within Unity along with position tracking, through a system of infrared cameras. To validate the performance of the real-time implementation, the orientation estimated from the algorithm and the position obtained from the infrared cameras are applied to a 3D hand model. An experiment requiring the acquisition of cubic targets within a 3D environment using the 3D hand motion tracking interface was performed 30 times. Experimental results show that the algorithm can be implemented in real-time and can eliminate the drift in orientation tracking.