Sensor Motion Tracking by IMM-Based Extended Kalman Filters

In this paper, we present a real-time motion estimation and tracking scheme using interacting multiple model (IMM) based Kalman filters. In the proposed IMM-based structure, two filters, quaternion-based extended Kalman filter (QBEKF) and gyroscope-based extended Kalman filter (GBEKF) are utilized for sensor motion state estimation. In the QBEKF, measurements from gyroscope, accelerometer and magnetometer are processed; while in the GBEKF, sole measurements from gyroscope are processed. The interacting multiple model algorithm is capable of adaptively fusing the estimated states from the two models, and generating better estimation results via the flexibility of model weighting. Simulation results validate the proposed estimator design concept, and show that the scheme is capable of reducing the overall estimation errors.