NavFormer: IGRF Forecasting in Moving Coordinate Frames

TL;DR

NavFormer introduces a rotation-invariant forecasting method for IGRF magnetic field targets using scalar features and a spectral stabilization module, achieving lower errors across diverse flight data.

Contribution

It presents a novel approach combining rotation-invariant features and a Canonical SPD module for improved IGRF forecasting in moving sensors.

Findings

Lower error than strong baselines in standard training

Effective in few shot training scenarios

Robust zero shot transfer performance

Abstract

Triad magnetometer components change with sensor attitude even when the IGRF total intensity target stays invariant. NavFormer forecasts this invariant target with rotation invariant scalar features and a Canonical SPD module that stabilizes the spectrum of window level second moments of the triads without sign discontinuities. The module builds a canonical frame from a Gram matrix per window and applies state dependent spectral scaling in the original coordinates. Experiments across five flights show lower error than strong baselines in standard training, few shot training, and zero shot transfer. The code is available at: https://anonymous.4open.science/r/NavFormer-Robust-IGRF-Forecasting-for-Autonomous-Navigators-0765