A weighted combination filter with nonholonomic constrains for integrated navigation systems

To meet the requirements of higher accuracy and stability of integrated navigation system, this paper applied Sage-Husa adaptive Kalman filter with nonholonomic constraints and forward/backward filtering to IMU/GPS integrated system, and the results of the forward and backward filtering are weighted and combined. A weighted combination filter is proposed in this paper, and which has been used in post-processing to improve MEMS IMU/GPS accuracy. Through the car navigation experiment, data set has been processed by four filtering algorithms. By means of comparing the four results, the method proposed for the vehicle integrated navigation system achieved the best accuracy with standard deviations of latitude = 1.03 m, longitude = 1.31 m, and heading angle = 0.84 deg°, which demonstrated the advantages of the new method.