Artificial Intelligence takes control: The first “self-driving” telescope successfully observes the night sky

AI has already transformed fields ranging from healthcare and finance to logistics and self-driving vehicles. Now, artificial intelligence is turning its attention to one of humanity’s oldest scientific endeavors: stargazing.

Researchers from Northwestern UniversityThe University of Chicago and the US Department of Energy’s Fermilab have successfully demonstrated what they describe as the first artificial intelligence-based telescope scheduling system, marking an important step toward autonomous astronomical observatories. This technology has already been tested at one of the world’s most productive astronomical facilities, demonstrating that AI can make complex observing decisions in real time while adapting to changing conditions throughout the night.

This breakthrough could help astronomers make better use of scarce telescope time, accelerate scientific discoveries, and provide valuable lessons for Canada’s fast-growing artificial intelligence and astronomy sectors.

The problem with traditional telescope scheduling

Observing the universe is much more complex than simply pointing a telescope at an interesting celestial body. Each night, astronomers must weigh multiple factors before deciding where to focus the telescope. Cloud cover, atmospheric stability, moonlight brightness, object visibility, science priorities, and telescope availability all play a role in determining the most productive use of observing time.

The challenge is particularly important because access to large telescopes is highly competitive. Researchers often wait months or even years to obtain approved observation periods. If conditions deteriorate or the telescope is aimed ineffectively, valuable scientific opportunities could be lost.

As Alex Drlica Wagner, a professor of astronomy and astrophysics at the University of Chicago and a scientist at Fermilab, notes, The major telescopes are international scientific resources It is used by researchers all over the world. Every minute of observation time holds great value.

Historically, these scheduling decisions have relied heavily on the expertise of experienced astronomers. The new artificial intelligence system aims to automate much of this process.

Training AI to think like an astronomer

Instead of programming the system with hundreds of specific rules developed over decades of telescope operations, the research team chose a different approach. They let the AI ​​learn by example.

The researchers trained a deep learning model using 13 years of collected historical observations By scanning dark energya large astronomy project using the Dark Energy Camera (DECam) mounted on the 4-meter Víctor M. Blanco telescope in Chile.

The system was shown where the telescope was observing at a given moment and then asked to predict its next target. By repeatedly comparing forecasts with decisions made by human astronomers, the AI ​​has learned the observing strategies that experts use when making scheduling choices.

According to the researchers, the system was not explicitly taught rules regarding moonlight conditions, atmospheric quality, or image enhancement. Instead, I learned these relationships independently of historical data.

This approach represents a growing trend in artificial intelligence: allowing machine learning systems to discover complex patterns that would otherwise be difficult to encode manually.

The real test came when the AI ​​was deployed in an operational observatory. Using the Dark Energy Camera, a state-of-the-art 570-megapixel instrument, researchers conducted two successful observations during the spring and summer of 2026 at the NSF Víctor M. Blanco Telescope at the Cerro Tololo Inter-American Observatory in Chile.

The AI ​​system created monitoring plans and adapted those plans as conditions evolved throughout the night. Changes in weather or sky conditions that may require human intervention are automatically incorporated into the revised observing schedules. For the initial deployment, the goal was modest but important: matching human performance.

According to the research team, the AI ​​scheduling software achieved results Similar to that of experienced human operators. Having demonstrated this capability, the next phase of development became considerably more ambitious.

Researchers now hope to create systems capable of outperforming human schedules by identifying monitoring strategies that people may never have thought of.

Modern astronomy is entering an unprecedented era of data generation. Next-generation facilities, such as the NSF-DOE Vera C. Rubin Observatory, are expected to produce vast amounts of astronomical data. Managing feedback and coordinating follow-up investigations will become increasingly difficult. AI may be uniquely suited to this environment.

Intelligent scheduling systems can address rapidly changing conditions, evaluate competing priorities, and optimize telescope utilization in ways that are difficult for human operators to do consistently over long periods. The result could be more efficient scientific processes and increased research productivity.

Importantly, automation can also free astronomers from routine operational decisions, allowing them to focus more attention on scientific interpretation and discovery.

As Drlica Wagner suggests, removing some of the technical burden of planning observations may enable researchers to spend more time addressing basic scientific questions.

Canadian point of view

Although the project was implemented in the United States and Chile, its implications are relevant to Canada. Canada has emerged as one of the world’s leading centers for AI research, thanks in large part to organizations like MILA in Montreal, the Vector Institute in Toronto, and AMIE in Edmonton. Canadian researchers played a major role in developing machine learning techniques now used around the world.

Canada also has a distinguished tradition in astronomy and astrophysics. Canadian scientists contribute to international telescope projects, cosmology research, exoplanet studies, and observational astronomy initiatives around the world.

The convergence of artificial intelligence and astronomy represents a particularly exciting opportunity. Projects like the Sky Artificial Intelligence (SkAI) Institute, which has supported telescope scheduling work, demonstrate how machine learning can become an active partner in scientific discovery. Similar methods could eventually be applied to observatories used by Canadian researchers or incorporated into future international collaborations in astronomy involving Canadian institutions.

There are also broader industrial impacts. The technologies developed for autonomous telescope operations are similar to the challenges faced by autonomous vehicles, intelligent manufacturing systems, robotics, and remote sensing platforms. Progress in one sector often generates innovations that benefit many other sectors.

The successful deployment of an AI scheduling system points to a future in which observatories become increasingly autonomous. Instead of relying on constant human supervision, future telescopes may monitor environmental conditions, select optimal targets, coordinate with other observatories, and automatically adjust observing strategies. Such systems may become especially important in remote environments where recruitment is difficult or expensive. They may also be necessary as the number of astronomical surveys continues to grow and observation becomes more complex.