Uncertainty-Based IMU Orientation Tracking Algorithm for Dynamic Motions

With the recent technological advancement in low-cost wireless inertial motion trackers, measuring three-dimensional motion for biomechanics research becomes more facile. However, the methods of acceleration modeling in off-the-shelf filters do not hold for all movements in sports activities with significant and long-lasting accelerations. This paper presents a robust algorithm for orientation tracking in the presence of large active accelerations lasting longer than the maximum time the MEMS gyroscopes can solely keep track of the body orientation. We particularly model the uncertainty of active acceleration and take it into explicit account in an extended Kalman filter based orientation estimator for applying measurement updates accurately in dynamic motions such as sports activities. The proposed tracker also estimates the magnetic disturbances by using an uncertainty model to improve the heading estimation. Benchmarking the results with the Vicon Optical as ground truth and the MTw kit with a specific filter for body motion tracking shows the robustness of our method against variations of acceleration in different types of motion. Our tracker performs orientation estimation in real time with fast convergence during acceleration shocks and low root-mean-square error, particularly when experiencing large accelerations in periodic motions.