A robust estimation method of GNSS/IMU fusion kalman filter

Aiming at the traditional gross error detection method of GNSS/IMU fusion Kalman filter, which simply retains or eliminates abnormal observation data and does not make full use of useful observation information, a new robust estimation method of Kalman filter is proposed. Based on the chi-square statistical hypothesis test, this method constructs continuous change measurement using weight coefficient, fully excavates measurable innovation between normal value and abnormal value, and establishes a new measurement update equation of Kalman filter robust estimation. Finally, through GNSS/IMU integrated navigation simulation, the advantages of the new Kalman filter are verified: no parameter adjustment is required, and the statistical error fluctuation is smaller than that of the traditional adaptive filter.