面向复杂路面环境的 SINS/LDV 鲁棒标定算法

To overcome the degradation of calibration accuracy caused by outliers and velocity measurement failures of laser Doppler velocimeters (LDV) on complex road surfaces such as water or mud, a robust calibration algorithm for Strapdown Inertial Navigation System (SINS)/LDV in complex road environment is proposed. Firstly, displacement increment vectors are constructed in a sliding window from SINS/GNSS and LDV data, and an exponentially weighted moving average method is adopted to enhance noise suppression, thereby improving the calibration accuracy of the LDV scaling factor. Then, by improving the robust vectorized K-matrix Kalman filter, noise in the observed vector is reduced, and the calibration accuracy of the installation error angles is improved. Simulations and vehicle experiments shows that the proposed algorithm successfully achieves relevant error parameter calibration even during LDV velocity measurement failures. Vehicle dead reckoning based on calibration results demonstrates that compared to the quaternion-based Kalman filtering and the gradient-descent quaternion algorithm, the proposed algorithm reduces horizontal positioning root mean square error (RMSE) by 18.18% and 28.74%, and altitude positioning RMSE by 22.68% and 45.69%, respectively. Over 177 km, the final horizontal positioning accuracy reaches 0.018%D, thereby enhancing the algorithm's robustness and calibration precision in complex road environment.