Reinforcement Learning Improves Wavefront Control for Exoplanet Imaging

This paper introduces a novel approach to focal plane wavefront control using model-based reinforcement learning. The algorithm, named Policy Optimization for NCPAs (PO4NCPA), addresses the critical challenge of correcting non-common-path aberrations (NCPA) in high-contrast imaging instruments. NCPA correction is essential for directly imaging exoplanets, particularly those orbiting close to their host stars. Conventional methods often rely on mechanical mirror probes, which can compromise performance. PO4NCPA overcomes this limitation by leveraging sequential phase diversity and interpreting the focal-plane image as input data. The algorithm then determines phase corrections that optimize both non-coronagraphic and post-coronagraphic point spread functions (PSFs) without requiring prior system knowledge. The simulations demonstrate that PO4NCPA effectively compensates for both static and dynamic NCPAs, achieving near-optimal focal-plane light suppression with a coronagraph and near-optimal Strehl without one. Its effectiveness under photon and background noise, combined with its sub-millisecond inference times, makes it a promising solution for real-time low-order correction of atmospheric turbulence beyond high-contrast imaging. This advancement could significantly enhance the capabilities of extremely large telescopes and facilitate the direct imaging of potentially habitable exoplanets.

_Context: This intelligence report was compiled by the DailyOrbitalWire Strategy Engine. Verified for Art. 50 Compliance._