China's AI-powered telescope selected as global AI application case

A telescope developed by the National Astronomical Observatories of China (NAOC) under the Chinese Academy of Sciences has been selected as a representative case of artificial intelligence (AI) applications in physics, astronomy, chemistry and materials science in Stanford University's 2026 AI Index Report.

The Xingyu Telescope was jointly developed by the Milky Way 3D Structure Group of the NAOC and the Xinglong Observational Base. It is a new astronomical embodied-intelligence observation system.

Unlike traditional applications of AI, which are mainly used for analyzing astronomical data after observations, the Xingyu Telescope deeply integrates large models and AI agents with the telescope's control system. This allows AI to participate throughout the astronomical observation process, changing the traditional approach to astronomical research.

Research-grade telescopes have complex structures, while observation performance can be affected by multiple factors, including equipment status, weather conditions and sensor parameters. Traditional observations therefore rely heavily on manual control.

To address this challenge, the research team developed a low-cost digital simulation system for telescopes.

The system can simulate different observation scenarios, test control strategies and train AI agents without taking up time on actual observations. The overall development cost can be kept at the thousand-yuan level (1,000 yuan equals roughly $149), significantly lowering the threshold for developing intelligent astronomical equipment and introducing a new approach to developing simulation systems for scientific research facilities.

The intelligent system has so far successfully issued early warnings for eight extremely early-stage supernova candidates, two of which have completed follow-up observations.

The AI agent can also automatically retain experience gained from observations and iteratively optimize observation strategies, continuously improving observation accuracy.