A unified ECEF-frame-based adaptive robust navigation method for all-latitude scenarios

As the strategic importance of polar regions continues to grow, there is an urgent demand for high-reliability navigation systems with globally unified modeling and robust filtering capabilities. To address the challenges encountered by global integrated navigation systems, including polar geometric singularities, degraded estimation consistency, and interference from measurement outliers, this paper proposes a unified ECEF-frame-based adaptive robust navigation method for all-latitude scenarios. First, a globally unified SINS/GNSS integrated navigation model is established to eliminate the error risks associated with navigation-frame switching. Second, in the ECEF frame, an improved integrated navigation model is derived in detail based on a rigorous definition of the velocity error; by removing the dependence of the system matrix on the specific-force term, the covariance-inconsistency issue is thereby mitigated. Third, a switching filtering strategy driven by measurement-outlier detection is developed, where a selective reweighting mechanism suppresses abnormal inflation of the prior covariance and ensures filtering robustness. Finally, an adaptive kernel-bandwidth regulation method based on the relative-efficiency criterion is developed, effectively reducing reliance on exhaustive manual parameter tuning. Experiments based on an open-source SINS/GNSS integrated navigation dataset and virtual latitude transformation evaluate the proposed algorithm under representative low-, mid-, and high-latitude scenarios.