We investigate the adoption of Machine Learning techniques for piston sensing in the context of segmented primary mirror telescopes by the means of numerical simulations. Considering a Natural Guide Star Wavefront Sensor, composed by one high order modes sensor plus a second sensor dedicated to the differential piston modes, we focus on the latter and tackle the problem of providing an accurate estimation for the piston modes coefficients from a defocused image of the system PSF.

We consider as a baseline algorithm a customized version of LIFT (which is based on a Maximum Likelihood Estimation) and compare its performance with a Deep Neural Network (DNN) regression. After considering several DNN architectures, we designed a simple one and performed some degree of hyperparameter optimization on it to obtain the final DNN version.

MORFEO/MAORY is the post-focal adaptive optics instrument of the ELT. It is designed to provide the 53×53 arcsec field of view of MICADO with MCAO correction based on split-tomography, where the Low-Order modes are sensed by three NGS-based WFS. To maximize the sky-coverage the LO-WFS are 2×2 subapertures Shack- Hartmann sensors working in the H band, making use of the FREDA detectors. MAORY also implements 3 dedicated NGS-based truth sensors to measure at slow rate the true higher order atmospheric aberrations and to de-trend the LGS WFS measurements. These WFS work with the visible light of the NGS to feed a 10 × 10 SH sensor that makes use of the ALICE detector. Each unit of LOR WFS is provided with a couple of orthogonal linear stages to allow for the NGS acquisition in a 80 arcsec radius. The 3 LOR WFS are arranged at 120° geometry on a common support structure that rigidly connects them to MICADO and its rotator.

We present the error budget of an astronomical single conjugated adaptive optics system with a pyramid wavefront sensor. The chosen approach is semianalytical to take into account the full response characteristics of this kind of sensor, and, in addition to these characteristics, it includes the closed-loop dynamics and all the main error sources: fitting, temporal, measurement, and aliasing error. This error budget has been made to complement more complex numerical simulators in the system design process, to rapidly explore a huge range of system parameters and to assess the critical components. Its reliability is demonstrated with a comparison with Monte Carlo numerical simulations on a reference system, the Single Conjugated Adaptive Optics Upgrade for Large Binocular Telescope.