Brommer, Christian (University Klagenfurt), Boehm, Christoph (University), Steinbrener, Jan (Universität Klagenfurt), Brockers, Roland (Jet Propulsion Laboratory, California Institute of Technology), Weiss, Stephan (University of Klagenfurt)

Improved State Estimation in Distorted Magnetic Fields

Scheduled for presentation during the Regular Session "Sensor Fusion" (ThC1), Thursday, September 3, 2020, 18:00−18:20, Macedonia Hall

2020 International Conference on Unmanned Aircraft Systems (ICUAS), September 1-4, 2020 (Postponed from June 9-12, 2020), Athens, Greece

This information is tentative and subject to change. Compiled on April 19, 2024

Abstract

The magnetic field of the Earth - besides varying naturally - may be disturbed locally by the (man-made) environment. State estimation and corresponding navigation frameworks that use magnetometers on-board of mobile platforms suffer from severe performance loss or failure upon local magnetic distortions if these are not detected and mitigated adequately. Advanced estimators include the magnetic variation in the state vector. However, Cartesian coordinates, although widely used in the literature, suffer from observability issues when it comes to disturbance detection. This paper shows the importance of representing the magnetic variation in a spherical coordinate system and subsequent improvements with respect to the common representation in Cartesian coordinates. The spherical representation improves estimator consistency and allows for accurate and fast mitigation of magnetic disturbances through consistent statistical tests which leads to better system state estimates in magnetically distorted areas. The approach is validated by performing tests with simulated and real-world data on embedded hardware.