Here we describe the continuation of our work on the installation of a control law on HOMER. Previously, we had tested an integrator type control law and the results seemed to be very promising. We then set up the Linear Quadratic Gaussian (LQG) control. This type of control is used to estimate the turbulence in the volume then its correction in real time by projection on the mirrors. It is based on the use of a Kalman filter for the estimation and prediction of the turbulence in the volume. It has a real value in the context of tomography AO tests that require a knowledge of the turbulent volume. This work on the LQG control was carried out in cooperation with the L2TI.
The turbulence conditions are slightly different compared to the first results presented. Effectively, we increased the conjugation altitude of DM2 to increase the anisoplanatism effect and the footrprint separation between the off-axis pupils at altitude. The turbulence was still equally distributed between the two turbulent layers. We studied the performance, always quantified with the value of the Strehl ratio on the star, in four directions: an on-axis star at the center of the field and 3 off-axis stars, that were used as multi-analysis guide stars.
We compared the results obtained in open loop (figure 1), in conventional AO with an integrator type control (figure 2), in GLAO [2] with an integrator control (figure 3), in tomography AO (figure 4) with an LQG control and in MCAO [3] with an integrator control (figure 5).
A TAO (Tomography Adaptive Optics) type correction analyzes in several directions to reconstruct the turbulent volume, then the turbulence is projected in a direction of interest. The correction is therefore made using a single pupil conjugated mirror. So this is an optical system similar to GLAO (same number of stars in the analysis and same number of DMs) however the method of reconstruction of the wave front is different and so therefore is the control.
Figure 1: Open loop long exposure image on HOMER in the presence of turbulence.
.jpg)
Figure 2: Closed loop long exposure image in conventional AO. The star circled in green corresponds to the on-axis (direction of correction) guide star. We clearly see the effect of the anisoplanatism that degrades the correction in the field.
Figure 3: Long exposure image of a GLAO type correction with an integrator control. The guide stars are circled red. The correction is uniform in the field but weak.
These experimental results confirm the respective advantages of the various concepts:
- AO gives a very good correction in the analysis direction but is limited by anisoplanatism,
- GLAO gives a uniform correction in the field but is quite weak,
- TAO can be used to correct in a field direction other than that of the analysis directions by means of a simple optical concept,
- MCAO gives a very good correction over the whole field.
We have therefore been able to carry out a first experimental installation of the LQG control on HOMER. The initial results are very encouraging. We now have to install this control in MCAO to evaluate the gain compared to that of an integrator type control.
