CNN regression and rotation invariance improve magnetic indoor localization

Researchers have developed a new indoor positioning system using convolutional neural networks (CNNs) and magnetic field data. This system addresses the challenge of device orientation sensitivity by employing rotation-invariant features derived from the magnetic field. The proposed model, MagNetS/XL, achieves state-of-the-art accuracy on the MagPie dataset, outperforming previous methods by maintaining accuracy even with significant device rotations. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT This research offers a more robust and infrastructure-free solution for indoor positioning, potentially improving applications in robotics and IoT.

RANK_REASON This is a research paper detailing a new method for indoor localization using magnetic fields and CNNs.

Read on arXiv cs.LG →

• CNN
• MagNetS/XL
• MagPie dataset
• arXiv

• paper
• other