ABSTRACT
We report on the use of artificial neural networks to estimate phase distortion in astronomical telescopes using focused images of a stellar source or an artificial laser guide star. The method was first developed as a means of measuring distortion induced by atmospheric turbulence and controlling an adaptive optics system for compensation of this atmospheric aberration. The method was then extended for use as a means of estimating static aberrations in the Hubble Space Telescope. We have tested the neural network aberration estimates against wavefront measurements of a Hartmann sensor, one of the traditional means of aberration measurement in adaptive optics systems, and have found good agreement. We have also compared the neural network with traditional high-resolution phase-retrieval methods with good agreement. The neural network approach offers a simple inexpensive way to implement adaptive optics in astronomical telescopes. It can also provide a quick and easy diagnostic tool for astronomical telescopes by providing estimates of static aberrations without any modification or disassembly of the telescope.
