A Fast Quaternion-Based Orientation Optimizer via Virtual Rotation for Human Motion Tracking

Abstract

For real-time ambulatory human motion tracking with low-cost inertial/magnetic sensors, a computationally efficient and robust algorithm for estimating orientation is critical. This paper presents a quaternion-based orientation optimizer for tracking human body motion, using triaxis rate gyro, accelerometer, and magnetometer signals. The proposed optimizer uses a Gauss-Newton (G-N) method for finding the best-fit quaternion. In order to decrease the computing time, the optimizer is formulated using a virtual rotation concept that allows very fast quaternion updates compared to the conventional G-N method. In addition, to guard against the effects of fast body motions and temporary ferromagnetic disturbances, a situational measurement vector selection procedure is adopted in conjunction with the G-N optimizer. The accuracy of orientation estimates is validated experimentally, using arm motion trials.