Roscosmos Integrates AI Into the Control of a New Orbital Station
Technology is reaching ever higher. Soon, largely autonomous systems will venture into space. Roscosmos plans to integrate artificial intelligence into the key control systems of the Russian Orbital Station (ROS). The move is driven by the complexity of future missions and the goal of reducing the workload on crews.
AI will automate equipment control, analyze telemetry and optimize the operation of onboard systems. The station is expected to learn to “think” for itself, making some technical decisions without constantly waiting for commands from Earth.
AI in Orbit
On a space station, everything is done in response to commands from Earth. Every step is monitored by the Mission Control Center. That was the old model. Now the station is becoming more like a smart assistant that can understand what is happening and make some decisions on its own.
Artificial intelligence monitors equipment and predicts failures rather than simply responding after something breaks. This can significantly reduce the amount of routine work astronauts have to do, including processing data, preparing reports and checking systems. Instead of spending their time at control consoles, they can focus on science and more complex tasks. If a signal is lost or delayed, the station can also continue operating. Machine-learning algorithms will increase the station’s autonomy and, in turn, the safety of long-duration missions, when communication delays or emergencies could pose a serious risk.
It is worth noting that the Russian Academy of Sciences announced back in 2021 that robotic systems and AI would be used aboard the future ROS. By 2025, Russian researchers had begun testing the GigaChat language model aboard the International Space Station as an assistant for astronauts. The neural network helps process voice recordings of experiments and prepare structured reports for transmission to Earth. According to Roscosmos, this has reduced the complexity of some procedures by about 20%. The system was specifically trained to work in difficult conditions: it can recognize speech against continuous background noise of 60–80 dB, understand professional slang, technical terms and abbreviations, and correctly process mixed Russian-English speech.
Technology Meets Space
In 2026, Roscosmos put about 60 spacecraft into orbit, bringing Russia’s overall constellation to nearly 400 satellites and other spacecraft. AI can analyze their imagery for almost everything from deforestation to urban growth. Continuous monitoring of Earth’s surface will soon be carried out by 132 small satellites in the Gryphon constellation. Terra Tech, Roscosmos’s center of expertise for AI applied to satellite data, has already created a library of more than 100 neural networks for processing imagery. Full-scale operation of the system is expected in 2026–2027.
AI is also moving directly onto spacecraft. Novosibirsk State University is developing the small NORBI-4 satellite, a 6U CubeSat. Its distinguishing feature is a neural module that will test AI functions directly in orbit. AI will handle information processing as well as algorithms for stabilizing and positioning the spacecraft.
TeleDroid, an anthropomorphic robot developed by Androidnaya Tekhnika and the successor to the well-known FEDOR robot, is also preparing for its first venture into open space this year. It is designed to work outside the station and will serve as the “eyes” and “hands” of the astronauts.
A Station That Works Like a Drone
The Russian Orbital Station is planned to be deployed between 2030 and 2034. “Artificial intelligence will already be fully integrated into the new orbital station we are building because it is unique as a drone station. It will be the first of its kind in the world. AI technologies will be used to allow the station to service itself,” said Dmitry Gorshkov, director of Roscosmos’s Digital Transformation Department.
Smart satellites and orbital systems will independently analyze information and respond to changes. AI will take on equipment diagnostics, failure prediction, station resource management and telemetry-stream analysis. And all of this will have to work under difficult conditions – amid radiation and without a constant internet connection. That means the technologies must be truly “space-grade.” What will Earth gain from them? Developing these technologies will make space services more efficient, from navigation to emergency monitoring. Space-based data will become part of everyday digital products.
