Orientation Correction for a 3D Hand Motion Tracking Interface Using Inertial Measurement Units

This paper outlines the use of an orientation correction algorithm for a miniature commercial-grade Inertial Measurement Unit to improve orientation tracking of human hand motion and also to improve 3D User Interfaces experience to become more realistic. The algorithm uses the combination of gyroscope, accelerometer and magnetometer measurements to eliminate the drift in orientation measurement which is caused by the accumulation of the bias offset error in the gyroscope readings. The algorithm consists of three parts, which are: (1) bias offset estimation, (2) quaternion correction using gravity vector and magnetic North vector, and (3) quaternion interpolation. The bias offset estimation is performed during periods when the sensor is estimated to be static, when the gyroscope reading would provide only the bias offset error for prediction. The quaternion was calculated based on unbiased angular velocity and then used to rotate the gravity vector and magnetic North vector in the Earth's frame resulting in the calculated gravity vector and magnetic North vector in the sensor's frame. The angular errors between calculated and measured gravity vector and the angle between calculated and measured magnetic North vector are used to calculate the correction quaternion that must be applied to the previous quaternion result. The result of the orientation estimation using this algorithm can be used to track the orientation of human hand motion with less drift and improved orientation accuracy than achieved with the on- board Kalman-based orientation filtering.