Physics Aware Neural Networks: Denoising for Magnetic Navigation

Aritra Das, Yashas Shende, Muskaan Chugh, Reva Laxmi Chauhan, Arghya Pathak, Debayan Gupta · 2026

Read Physics Aware Neural Networks: Denoising for Magnetic Navigation by Aritra Das, Yashas Shende, Muskaan Chugh, Reva Laxmi Chauhan, Arghya Pathak, Debayan Gupta (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-13690.txt.

Description

Magnetic navigation offers an alternative when GPS is out or broken, but an aircraft's own magnetism corrupts the field signal, and the old Tolles-Lawson model handles random noise poorly. This paper imposes two physics constraints. Divergence-free vector fields are built by learning a vector potential whose curl defines the field, and E(3) symmetry is built from tensor products of geometric tensors via sphere functions. Together they enforce Maxwell's equations and correct shifts with sensor position and orientation.

Keeping the field true to physics restricts the function space, acting as a quiet guard that improves time-space performance, with the Contiformer, combining continuous-time dynamics and memory. Made-up data from the World Magnetic Model and time-series GANs eases scarcity. Experiments show embedding these constraints greatly improves forecast accuracy and physical plausibility, beating old and unconstrained deep learning approaches. Aritra Das leads the work with five co-authors.

Catalog record

CREATORAritra Das, Yashas Shende, Muskaan Chugh, Reva Laxmi Chauhan, Arghya Pathak, Debayan Gupta
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-13690

Browse the full library · About the Tower of Babel