Polarization-Based Underwater Geolocalization with Deep Learning
2 institutional releases
Published in eLight · View the paper (DOI)
Researchers developed a novel method for underwater geolocalization using deep neural networks trained on 10 million polarization-sensitive images. The technology enables tethered-free navigation and has the potential to improve location accuracy, enabling in situ autonomous sampling robots to monitor water properties.
Coverage from 2 institutions
- Navigating the future of underwater geolocalization: how polarization patterns enable new technology University of Illinois Grainger College of Engineering · Jul 10, 2023 · first to report
- Navigating underwater inspired by migratory animals Light Publishing Center, Changchun Institute of Optics, Fine Mechanics And Physics, CAS · Jul 11, 2023