First results of an on-line adaptive optics system with atmospheric wavefront sensing by an artificial neural network

We report the first results from an adaptive optics system operating on-line at the telescope with the wavefront aberration sensed by a trained artificial neural network. Star images were formed at 2.2 μn wavelength by two coherently phased apertures of the Multiple Mirror Telescope (MMT), and analyzed by the neural net. The net derives wavefront parameters in a few milliseconds, and the system performance is fast enough that the aberration is nearly frozen during the time needed to make a correction. With the servo loop in operation, the corrected image shows significant power at the diffraction limit of 0″.1. We discuss some areas of scientific interest which will benefit from the application of adaptive wavefront correction.