Quaternion based constrained algorithm in federated Kalman filtering for MEMS IMU/GNSS

The performance of MEMS IMU/GNSS integrated system would degrade with outages of GNSS signal. For land vehicle applications, a federated Kalman filter was established. The quaternion was employed as part of states in one of the local filters. The acceleration constraint derived from quaternion was applied along the vehicle's body frame to expand observables in the filter. The field test shows that, compared with conventional filtering algorithms, the proposed algorithm brings 25% performance improvement in three-dimension(3D) positioning during 30 s GNSS outages. The accuracy of attitude and velocity is improved as well.