Abstract
To recover the resolution lost in a ground-based telescopes due to the atmospheric turbulence, it is necessary to use a technique known as Adaptive Optics (AO). The next generation of telescopes will have primary mirrors of more than 30 meter in diameter and will require AO systems from the ground up (Nelson et al. 2006). There are a number of challenges to implement an AO system at these scales. One of these challenges is the accurate measurement of the aberrated wavefronts using a laser guide star and a Shack-Hartmann wavefront sensor. Due to the diameter of the telescope and the use of the sodium layer in the upper atmosphere as photon return for the laser guide stars, the image of the guide star will appear elongated in the wavefront sensor. Typical centroiding algorithms such as Center of Gravity do not perform well under these conditions (Thomas et al. 2008). We present a new technique based on artificial neural networks for measuring the spot position with better accuracy than existing methods. Simulation results confirms that the new algorithm incurs in smaller errors with respect to other centroiding techniques in use.
