Answers are grounded in this paper and cite their sources.
Geophysical map-matching navigation becomes unreliable where the local field gradient vanishes, a limitation inherent to matching against a single scalar map. Gravity and magnetic anomalies depend on different rock properties (density and magnetization) and occupy different wavelength bands, so their gradients are largely uncorrelated (Pearson r ≈ 0.03 r ≈ 0.03 on the Korean maps used here). We exploit this complementarity by fusing an optically pumped magnetometer and a cold-atom gravimeter with an error-state INS, and compare centralized and federated EKF architectures on Korean national geophysical databases, with measurements simulated from the same maps the filters match against. In a 100-seed Monte Carlo with a tactical-grade IMU, the dual EKF reduces the RMS position error relative to MagNav-only by 86–90% on the gravity-strong and intermittent-magnetic paths and by 62% on the magnetic-strong path, and remains bounded on intermittent-magnetic terrain, where the single-map EKF diverges. A marginalized particle-filter control on that path attributes much of this gap to EKF linearization and shows that the second map mainly lowers the rate of large failures. With a navigation-grade IMU the dual advantage is significant under idealized noise but is lost under a realistic noise model combining correlated gravimeter noise and magnetic compensation residuals. A single-run gravimeter noise sweep on a representative path places the onset of the navigation-grade dual benefit near 0.5 mGal, with the full benefit under idealized noise at 0.1 mGal; the tactical-grade benefit persists across the sweep.
No takes yet. Share an insight, caveat, or question.
Lee et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: