An INS/GNSS integrated Navigation Algorithm Based on Robust Adaptive Estimation

In order to overcome the limitation of federal Kalman filter in multi-sensor information fusion when the statistical information of system noise and measurement noise is not accurate, a new federal filter algorithm based on robust adaptive estimation was proposed. Prediction residual statistics was used to determine abnormal observations on the base of reliable system model. When the exception was detected in a certain degree of confidence, the corresponding covariance matrix of observations was expanded. So that the filter has the property of robustness. INS/GNSS integrated navigation system experiments show that the new federal filter algorithm based on robust adaptive estimation is prior than the traditional federal Kalman filter.