Optimization of Closed-Loop Adaptive-Optics Control Algorithms Using Measured Performance Data: Experimental Results

Experiments have shown the reward adaptive-optics provides in improving the resolution of ground-based astronomical telescopes [1,2,3]. A critical contributor to adaptive-optics system performance is the control algorithm that converts wavefront sensor (WFS) measurements into the deformable mirror (DM) actuator commands. For the adaptive-optics systems in use today this control algorithm consists of a wavefront reconstruction step to estimate the instantaneous phase distortion to be compensated [4], followed by a servo control law to temporally filter this instantaneous estimate before it is applied to the deformable mirror [5]. So-called modal adaptive-optics systems can apply different temporal filters to separate spatial components, or modes, of the overall phase distortion [6]. Extensive analysis has been performed to evaluate and optimize the performance of these adaptive-optics control systems [7,8,9,10,11], but the results obtained depend on atmospheric parameters which are seldom known exactly and are constantly fluctuating. The uncertainty and variability of atmospheric conditions implies that an optimal degree of turbulence compensation cannot be achieved or maintained for long time intervals with a fixed control algorithm. A need exists for methods to update adaptive-optics control algorithms based upon actual system performance. Encouraging results have already been obtained demonstrating the value of emperically optimizing the control bandwidths for a modal adaptive-optics system [12]. In comparison, the subject of real-time adjustments to reconstruction matrices on the basis of measured system performance has received little attention.