Abstract

Open-loop adaptive optics is a technique in which the turbulent wavefront is measured before it hits the deformable mirror for correction; therefore the correct control of the mirror in open-loop is key in achieving the expected level of correction. In this paper, we present non-parametric estimation techniques to model deformable mirrors working in open-loop. We have results with mirrors characterized by non-linear behavior: a Xinetics electrostrictive mirror and a Boston Micromachines MEMS mirror. The inputs for these models are the wavefront corrections to apply to the mirror and the outputs are the set of voltages to shape the mirror. We have performed experiments on both mirrors, achieving Go-To errors relative to peak-to-peak wavefront excursion in the order of 1 % RMS for the Xinetics mirror and 3 % RMS for the Boston mirror . These techniques are trained with interferometric data from the mirror under control; therefore they do not depend on the physical parameters of the device. © 2010 SPIE.