SkyPole: a geolocation algorithm based on polarized vision without using astronomical ephemerides - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2024

SkyPole: a geolocation algorithm based on polarized vision without using astronomical ephemerides

Résumé

Global Navigation Satellite Systems (GNSS) are widely used due to their easy access to outdoor GNSS signals and their spatial precision. However, such systems are sensitive to jamming and spoofing. Simple and robust navigation strategies can be found in animals deprived by essence of any GNSS system. Studies have shown that animals like bees or ants utilize the sky’s polarization pattern for navigation. We recently proposed a method inspired by migratory birds, which calibrate their magnetic compass through the celestial rotation of night stars or the daytime polarization pattern. By considering the temporal properties of the sky’s polarization pattern as a relevant navigation information, we developed a bio-inspired method to find the geographical north bearing and the observer’s latitude, requiring only skylight polarization observations during the day. To reduce the noise susceptibility of our method, we added a pre-processing step using a least square method based on skylight polarization models, and a segmentation process based on a convolutional autoencoder neural network, trained with simulated data.
Fichier principal
Vignette du fichier
2024-Kronland-Martinet et al. - SPIE Defense+Commercial Sensing 2024 - submitted.pdf (1.39 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04607780 , version 1 (11-06-2024)

Licence

Identifiants

Citer

Thomas Kronland-Martinet, Léo Poughon, Marcel Pasquinelli, David Duché, Julien Serres, et al.. SkyPole: a geolocation algorithm based on polarized vision without using astronomical ephemerides. Polarization: Measurement, Analysis, and Remote Sensing XVI, Apr 2024, National Harbor, Maryland, United States. pp.74-79, ⟨10.1117/12.3013560⟩. ⟨hal-04607780⟩
38 Consultations
5 Téléchargements

Altmetric

Partager

Gmail Mastodon Facebook X LinkedIn More