Ponente
Descripción
Atmospheric turbulence constitutes one of the main limitations for large ground-based telescopes. In both astronomy and free-space optical communications (FSOC), turbulence degrades wavefront quality and reduces the stability of optical links. This work presents the development and implementation of Machine Learning–based methodologies for wavefront reconstruction and compensation, together with their experimental validation on an adaptive optics (AO) optical bench.
The work combines numerical simulations, synthetic dataset generation, and laboratory experimentation using Shack–Hartmann (SH) sensors, deformable mirrors (DM), and spatial light modulators (SLM). The performance of different deep learning models applied to wavefront reconstruction and atmospheric turbulence prediction is studied, evaluating their capability to operate in real time under different turbulence conditions. In addition, the application of these techniques to free-space optical communication scenarios is explored.
The experimental bench enables the validation of the developed methodologies under controlled conditions and allows their performance to be compared against conventional wavefront reconstruction techniques.
The expected results will contribute to the development of more robust, scalable, and computationally efficient AO systems, facilitating their integration into future large-aperture telescopes and advanced optical communication systems.