Relatore
Descrizione
The synergy between astronomy and photonics is opening new perspectives in wavefront sensing for ground-based telescopes. In this context, a further integration of photonic devices with machine learning techniques represents a particularly promising approach.
This work investigates the use of multimode fibers (MMFs) as wavefront sensors, coupled with a properly trained Neural Network (NN). In an initial work phase, aberrated wavefronts were propagated through a simulated MMF with known characteristics, and the resulting intensity patterns were used to train a given network architecture, assessing its ability to recognize the input aberrations.
To experimentally validate the concept, an optical bench has been developed and operated under controlled laboratory conditions, following the same approach adopted in the simulation phase. In this presentation, the methodology and the results obtained so far will be discussed.
| Sessione | Calcolo, Archivi e Intelligenza Artificiale |
|---|