---
title: "AI Takes Control of the Telescope: How a Neural Network Optimizes Astronomical Observations"
description: "For the first time, AI fully controlled a major telescope in Chile, independently planning observations while accounting for weather and the Moon. The algorithm's efficiency has already matched human performance, which is critical for the coming era of petabyte-scale data from the Vera C. Rubin Observatory 🌌🔭🤖"
date: 2026-08-02T15:53:06.000Z
lang: en
url: https://xab.info/en/posts/ai-takes-control-of-the-telescope-how-a-neural-network-optimizes-astronomical-observations
tags: [artificial-intelligence, astronomy, fermilab, vera-rubin-observatory, dark-energy-survey]
publisher: "XAB.info"
---

# AI Takes Control of the Telescope: How a Neural Network Optimizes Astronomical Observations

![Astronomical observatory under a starry sky with the Milky Way — symbolizing AI takeover of telescope control for optimized celestial observations](https://xab.info/media/2026/08/02/ii-uzal-upravlenie-teleskopom-na-sebe-kak-nejroset-optimiziruet-astronomicheskie-nablyudenija/ii-uzal-upravlenie-teleskopom-na-sebe-kak-nejroset-optimiziruet-astronomicheskie-nablyudenija-1.webp)

A new era has dawned in astronomy: for the first time in history, an artificial intelligence system has taken full control of observations at a major national telescope. Scientists from Northwestern University, the University of Chicago, and Fermi National Accelerator Laboratory have demonstrated that deep learning algorithms can make complex real-time decisions comparable to the experience of professional astronomers.

### Algorithm vs. Weather and the Moon

The tests were conducted at the observatory in Chile using the 4-meter Víctor M. Blanco telescope. The system controlled the 570-megapixel Dark Energy Camera (DECam). The task facing the AI was non-trivial: it had to independently determine where to point the instrument and instantly adjust the observation plan.

Unlike static programs, the algorithm accounted for dynamic factors: changing cloud cover, moon brightness, and current atmospheric conditions. Usually, such decisions are made by humans, who are forced to find a compromise between weather conditions, limited time, and the scientific value of targets to maximize the efficiency of expensive telescope operating hours.

### Learning from the Experience of Predecessors

The developers chose an unconventional path to train the model. Instead of manually coding the rules accumulated by astronomers over decades, researchers trained the neural network on the observation archive of the Dark Energy Survey project.

The model studied sequences of past observations, attempting to predict the next object in the list. The algorithm's decisions were compared with the actual actions of specialists, and errors became the basis for retraining. As a result, the system learned to intuitively account for the influence of the lunar phase and atmospheric conditions without explicitly programming these dependencies.

### Efficiency and Prospects

In the spring and summer of this year, the system successfully completed two full observation campaigns. According to the developers, the AI's efficiency is already comparable to human performance. However, the goal of the project is not limited to automating routine actions.

The next stage is to teach the model to find more efficient planning strategies that a human might not notice. This is particularly relevant on the eve of the commissioning of the Vera C. Rubin Observatory. The new giant will collect unprecedented petabytes of data daily.

The project authors are confident: intelligent planning systems will help coordinate the work of multiple telescopes, optimize every night of observation, and free scientists from routine tasks, allowing them to focus on data analysis and fundamental discoveries.