Aiming at the influence of star sensor anomaly measurement on the state estimation performance of the combined system, an inertial/astronomical combined navigation filtering algorithm based on Huber robust estimation theory is proposed. Through the detailed derivation of generalized maximum likelihood estimation theory, the Kalman filter algorithm flow based on Huber robustness estimation is given, and the performance of the algorithm is compared and verified based on actual sports car data. The results show that the proposed algorithm has a good robustness to the noise of star sensor anomaly measurement. Compared with the traditional Kalman filtering algorithm, the estimation accuracy and stability of the gyro constant drift are improved, and the accuracy of the inertial/astronomical combined system is guaranteed.