Aiming at the problem of outliers and noise interference in GNSS (Global Navigation Satellite System) / INS (Inertial Navigation System) coupling navigation, this paper applies a precise fusion algorithm based on REKF (Robust Extended Kalman Filter) to achieve high robustness and high-precision navigation in complex environments, and conducts performance evaluation. The REKF method enhances the robustness of the traditional EKF (Extended Kalman Filter) to abnormal observations by applying a robustness mechanism. The state is predicted using the INS model, and the time is updated through the data of the accelerometer and gyroscope to obtain a preliminary state estimate. The observation is updated according to the GNSS observation data, and the residual between the predicted state and the actual observation is calculated. The robustness function is applied to weight the residual, and the importance of the observation data is dynamically adjusted. Based on the adjusted observation noise covariance matrix, the Kalman gain is calculated and the state estimate is updated to obtain the optimal system state. The experiment analyzes the performance of three filtering algorithms, EKF, UKF (Unscented Kalman Filter), and REKF, in different scenes, and designs high dynamic scenes and weak signal environments as test platforms. The results show that in 10 specific scenes, the navigation RMSE (Root Mean Square Error) of REKF is smaller than that of EKF and UKF, and the average run time of EKF, UKF, and REKF is 32.5ms, 37.0ms, and 29.1ms, respectively. REKF is superior to traditional EKF and UKF algorithms in terms of accuracy and real-time performance. In complex and high dynamic environments, REKF can effectively improve positioning accuracy and reduce positioning errors, which has strong application potential.