Research team of the Julius Maximilian University of Würzburg (JMU) successfully tested the satellite orientation controller in orbit, based on artificial intelligence.
This is the first such experiment in the world. Тестування відбулося на борту наносупутника InnoCube формату 3U.
30 October 2025 року під час проходу супутника агент ШІ, розроблений у JMU, здійснив повний маневр орієнтації на орбіті, fully controlled by artificial intelligence. Using flywheels, The AI guided the satellite from its current initial orientation to a given target orientation. У наступних тестах він також успішно та безпечно керував супутником до цільової орієнтації. Дослідницька команда LeLaR таким чином зробила вирішальний крок до космічної автономності.
Проєкт In-Orbit Demonstrator for Learning Attitude Control (LeLaR) aims to develop autonomous next-generation orientation control systems. The main focus is on development, training and on-orbit testing of the controller based on artificial intelligence installed on the InnoCube nanosatellite. Stabilization of satellites in orbit prevents their chaotic rotation. in addition, the orientation controllers are also used to point the spacecraft in the desired direction during shooting, or for better communication with a ground station.
The peculiarity of this work is that, that the Würzburg controller was not created using traditional, fixed algorithms. Instead, the researchers took a deep reinforcement learning approach (DRL) - the field of machine learning, in which a neural network autonomously learns the optimal control strategy in a simulated environment.
The key advantage of the DRL approach is its speed and flexibility in comparison
with classical development of control systems. Traditional orientation controllers often require extensive manual parameter tuning by engineers – sometimes taking months or even years. The DRL method automates this process. in addition, it opens the possibility to create controllers, which automatically adapt to differences between expected and actual conditions, eliminating the need for lengthy manual reconfiguration.
Before implementation, the AI controller was trained on Earth in a highly detailed simulation, and then loaded into a flight model of the satellite in orbit. One of the biggest challenges was bridging the so-called gap between simulation and reality (Sim2Real) - ensuring that, so that the controller, trained in simulation, also worked effectively on a real satellite in space.
Having successfully demonstrated an AI-based controller in orbit, the Wurzburg team showed, that artificial intelligence can be reliably applied in safety-critical space missions. This should greatly increase the acceptability of AI methods in the future
in aeronautics and space research. Growing confidence in such technology is a crucial step toward future autonomous missions, example, interplanetary or deep space missions, where human intervention is impossible due to vast distances or communication delays.
Source: https://universemagazine.com