Which Error Sources Dominate Hybrid Quantum–Classical Inertial Navigation? A Sobol’ Variance Decomposition Across Atmospheric GNSS-Denied and Cislunar Regimes
Ansh Pathak - 2026 - Preprint
A validated open-source simulator of a cold-atom interferometer + classical IMU hybrid navigator (Cheiney–Lautier–Wang architecture in a 15-state error-state EKF), paired with a global Sobol′ variance decomposition of its error budget across two opposite aerospace regimes: an atmospheric UAV under GNSS jamming, and a multi-day cislunar coast. Across 27,648 simulations and eight passing validation tests, the hybrid holds a median 7.6 m after five minutes of jamming (about 4.9× tighter than classical-only) and 124 km after a 4.98-day lunar coast versus 38,800 km unaided (313×). The study shows the remaining atmospheric error is dominated by gyroscope bias and platform vibration, while the cislunar floor is governed by classical accelerometer bias and coast duration-not cold-atom stability alone. The preprint is archived at DOI 10.5281/zenodo.21855434.
- 27,648 - Sobol′ simulations
- 4.9× - Tighter · 5-min GNSS jamming
- 313× - Tighter · 5-day cislunar coast
- 8/8 - Validation tests passed