Reinforcement learning controller successfully demonstrated on sky for adaptive optics

By PulseAugur Editorial · [1 sources] · 2026-06-09 12:26

Researchers have demonstrated a reinforcement learning controller for adaptive optics systems on sky for the first time. Named PO4AO, the system was deployed on a telescope and consistently outperformed a standard integrator controller. It showed robustness to noise and vibrations, operating effectively across various conditions with a single set of hyperparameters. AI

IMPACT Demonstrates the practical application of reinforcement learning in complex scientific instrumentation, potentially improving astronomical observation quality.

RANK_REASON This is a research paper detailing the first on-sky demonstration of a novel reinforcement learning controller for adaptive optics. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv cs.LG →

paper
other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

COVERAGE [1]

arXiv cs.LG TIER_1 English(EN) · Byron Engler · 2026-06-09 12:26

On-sky demonstration of reinforcement learning for adaptive optics control

Reinforcement learning (RL)-based algorithms have recently emerged as a promising approach for adaptive optics (AO) control. In simulations and laboratory experiments, they have demonstrated robustness to real-world effects such as photon and detector noise, misregistration, vibr…

COVERAGE [1]

On-sky demonstration of reinforcement learning for adaptive optics control

RELATED ENTITIES

RELATED TOPICS