Airborne navigation based on magnetic field anomalies : Evaluation of high fidelity maps and a marginalized particle filter

Sammanfattning

Developing complementary navigation alternatives to Global Navigation Satellite System (GNSS) is a critical challenge for flying vehicles. GNSS is vulnerable to signal jamming and spoofing, creating a need for robust backup systems when satellite signals are compromised or unavailable. A promising solution to this problem is using the Earth’s magnetic anomaly field for navigation. This will be the focus of this thesis. The objective of the thesis is to study how factors such as reference map resolution, flight altitude, filter parametrisation, Inertial Measurement Unit (IMU) sensor noise, and temporal magnetic variations influence the performance of magnetic anomaly based navigation along a flight path over northern Sweden. To evaluate this, a 12-state Marginalized Particle Filter (MPF) was implemented and analysed through Monte Carlo simulations. It was shown in this thesis that the map quality highly impacted the performance; using high-resolution regional data from the Geological Survey of Sweden (SGU) significantly improved positioning accuracy compared to using a global dataset. Furthermore, the results demonstrate a clear trade-off between altitude dependent spatial information density and filter stability. Lower altitudes yielded high positioning accuracy but challenged filter stability. Higher altitudes are more stable but yield lower positioning accuracy. The MPF demonstrated strong resilience against temporal magnetic variations and varying IMU noise levels, although severe IMU degradation eventually led to altitude error divergence. Ultimately, this thesis concludes that the reference map quality and the flight altitude are the most critical factors affecting positioning accuracy, whereas the filter itself provides robust performance against moderate environmental and sensor disturbances.