Advances in real-time control algorithms

Abstract

Since the early days, many aspects of Adaptive Optics (AO) have seen tremendous changes. From the early experimental systems providing low order correction in a tiny patch of sky to todays fully automated specialized system offering correction in a much wider field and/or a much higher degree of correction, the evolution has been remarkable. For example, deformable mirror (DM) technology and wavefront sensing methods have been constantly improved. As well, real-time control algorithms have been greatly refined. This paper will review the different real-time control strategies that have been used with astronomical adaptive optics. They all have in common the same objective, that is the derivation of an optimal command for the deformable mirror(s) in order to get the least amount of residual optical aberrations in the science path. Most of the time, the real-time control algorithm is split in two independent components, the first part performing the wavefront (spatial) reconstruction, the second part performing the temporal control. With the advent of the extremely large telescopes (ELTs), as well as new AO modalities requiring several DMs and wavefront sensors, performing both these tasks in an ever shrinking glimpse of time is even more challenging. We will describe advanced fast and iterative reconstruction methods recently proposed for next generation AO systems. We will show how these algorithms combined with sparse matrices and parallel computing techniques meet the requirements of Extremely Large Telescope (ELT) real time computers.