Making the unmodulated Pyramid wavefront sensor smart. Closed-loop demonstration of neural network wavefront reconstruction with MagAO-X

Almost all current and future high-contrast imaging instruments will use a Pyramid wavefront sensor (PWFS) as a primary or secondary wavefront sensor. The main issue with the PWFS is its nonlinear response to large phase aberrations, especially under strong atmospheric turbulence. Most instruments try to increase its linearity range by using dynamic modulation, but this leads to decreased sensitivity, most prominently for low-order modes, and makes it blind to petal-piston modes. In the push toward high-contrast imaging of fainter stars and deeper contrasts, there is a strong interest in using the PWFS in its unmodulated form. Here, we present closed-loop lab results of a nonlinear reconstructor for the unmodulated PWFS of the Magellan Adaptive Optics eXtreme (MagAO-X) system based on convolutional neural networks (CNNs). We show that our nonlinear reconstructor has a dynamic range of >600 nm root-mean-square (RMS), significantly outperforming the linear reconstructor that only has a 50 nm RMS dynamic range. The reconstructor behaves well in closed loop and can obtain >80% Strehl at 875 nm under a large variety of conditions and reaches higher Strehl ratios than the linear reconstructor under all simulated conditions. The CNN reconstructor also achieves the theoretical sensitivity limit of a PWFS, showing that it does not lose its sensitivity in exchange for dynamic range. The current CNN's computational time is 690 microseconds, which enables loop speeds of >1 kHz. On-sky tests are foreseen soon and will be important for pushing future high-contrast imaging instruments toward their limits.

Tempestas ex machina: A review of machine learning methods for wavefront control

As we look to the next generation of adaptive optics systems, now is the time to develop and explore the technologies that will allow us to image rocky Earth-like planets; wavefront control algorithms are not only a crucial component of these systems, but can benefit our adaptive optics systems without requiring increased detector speed and sensitivity or more effective and efficient deformable mirrors. To date, most observatories run the workhorse of their wavefront control as a classic integral controller, which estimates a correction from wavefront sensor residuals, and attempts to apply that correction as fast as possible in closed-loop. An integrator of this nature fails to address temporal lag errors that evolve over scales faster than the correction time, as well as vibrations or dynamic errors within the system that are not encapsulated in the wavefront sensor residuals; these errors impact high contrast imaging systems with complex coronagraphs. With the rise in popularity of machine learning, many are investigating applying modern machine learning methods to wavefront control. Furthermore, many linear implementations of machine learning methods (under varying aliases) have been in development for wavefront control for the last 30-odd years. With this work we define machine learning in its simplest terms, explore the most common machine learning methods applied in the context of this problem, and present a review of the literature concerning novel machine learning approaches to wavefront control.

Self-optimizing adaptive optics control with Reinforcement Learning for high-contrast imaging

Current and future high-contrast imaging instruments require extreme adaptive optics (XAO) systems to reach contrasts necessary to directly image exoplanets. Telescope vibrations and the temporal error induced by the latency of the control loop limit the performance of these systems. One way to reduce these effects is to use predictive control. We describe how model-free Reinforcement Learning can be used to optimize a Recurrent Neural Network controller for closed-loop predictive control. First, we verify our proposed approach for tip-tilt control in simulations and a lab setup. The results show that this algorithm can effectively learn to mitigate vibrations and reduce the residuals for power-law input turbulence as compared to an optimal gain integrator. We also show that the controller can learn to minimize random vibrations without requiring online updating of the control law. Next, we show in simulations that our algorithm can also be applied to the control of a high-order deformable mirror. We demonstrate that our controller can provide two orders of magnitude improvement in contrast at small separations under stationary turbulence. Furthermore, we show more than an order of magnitude improvement in contrast for different wind velocities and directions without requiring online updating of the control law.