Share Email Print
cover

Proceedings Paper

Tomographic reconstructor for multi-object adaptive optics using artificial neural networks
Author(s): Dani Guzman; Alexandre T. Mello; James Osborn; Francisco J. De Cos; Marlon Gómez; Timothy Butterley; Nicole David; Nieves Roqueñi; Richard M. Myers; Andrés Guesalaga; Matias Salas
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Multi-object adaptive optics requires a tomographic reconstructor to compute the AO correction for scientific targets within the field, using measurements of incoming turbulence from guide stars angularly separated from the science targets. We have developed a reconstructor using an artificial neural network, which is trained in simulation only. We obtained similar or better results than current reconstructors, such as least-squares and Learn and Apply, in simulation and also tested the new technique in the laboratory. The method is robust and can cope well with variations in the atmospheric conditions. We present the technique, our latest results and plans for a full MOAO experiment.

Paper Details

Date Published: 13 September 2012
PDF: 11 pages
Proc. SPIE 8447, Adaptive Optics Systems III, 844740 (13 September 2012); doi: 10.1117/12.925355
Show Author Affiliations
Dani Guzman, Pontificia Univ. Católica de Chile (Chile)
Alexandre T. Mello, Pontificia Univ. Católica de Chile (Chile)
Univ. Federal de Santa Catarina (Brazil)
James Osborn, Pontificia Univ. Católica de Chile (Chile)
Francisco J. De Cos, Univ. de Oviedo (Spain)
Marlon Gómez, Univ. de Oviedo (Spain)
Timothy Butterley, Durham Univ. (United Kingdom)
Nicole David, Pontificia Univ. Católica de Chile (Chile)
Nieves Roqueñi, Univ. de Oviedo (Spain)
Richard M. Myers, Durham Univ. (United Kingdom)
Andrés Guesalaga, Pontificia Univ. Católica de Chile (Chile)
Matias Salas, Pontificia Univ. Católica de Chile (Chile)


Published in SPIE Proceedings Vol. 8447:
Adaptive Optics Systems III
Brent L. Ellerbroek; Enrico Marchetti; Jean-Pierre Véran, Editor(s)

© SPIE. Terms of Use
Back to Top