Experimental study on modified linear quadratic Gaussian control for adaptive optics.

To achieve high-resolution imaging the standard control algorithm used for classical adaptive optics (AO) is the simple but efficient proportional-integral (PI) controller. The goal is to minimize the rms error of the residual wave front. However, using the PI controller, it is not possible to do this. One possible way to minimize the rms error is to use linear quadratic Gaussian (LQG) control. In practice, however, this control algorithm still encounters an unexpected problem that leads to the divergence of control in AO. This paper proposes a modified LQG (MLQG) to solve this issue. The controller is analyzed explicitly. Laboratory tests shows strong stability and high precision compared to the classical control.