Using AI for more reliable space missions


Enabling & Support

30/01/2025
478 views
9 likes

As we prepare to go deeper into space, the demand for autonomous systems capable of operating independently from ground control and crew interactions is increasing. Artificial Intelligence (AI) is shaping up to be an essential tool for reaching this goal. With support from ESA’s Discovery programme, a team of researchers from Airbus explored how AI can collect and analyse data onboard the Columbus module of the International Space Station (ISS) in order to improve its prognosis and fault detection capabilities.

The developed AI system demonstrator – ORBIT-STAR monitors telemetry data to detect and anticipate any issues within a Columbus subsystem. Using this information and set guidelines, it can independently – identify actions to prevent further damage. The AI demonstrator also keeps track of its own decisions to reduce errors and improve itself over time. Additionally, when detecting a fault, the relevant data is sent to Ground Control, to support further analysis and systems improvements.

ORBIT-STAR Demonstrator

“By using AI we can enhance current capabilities onboard the Columbus module, increase sensitivity and even introduce new capabilities. Besides testing AI models, we gain valuable information about how to integrate these new models into the existing Columbus system and how to communicate with the Columbus Control Centre (Col-CC) in Germany. This activity closes various knowledge gaps,” says Luis Mansilla Garcia, AI System Engineer and ESA lead on this activity.

Going further than the ISS

Robots in space – artificial intelligence and beyond

This very promising system could ensure the safety and success of long-term missions in unknown environments, being able to adapt to new challenges with minimal help from human operators. “We need this technology in space to go deeper into space, where there is no connection to the ground,” says Christoph Haskamp, AI Expert at Airbus Defence and Space GmbH.

Closer to home, AI can be a valuable tool for future applications in low Earth orbit (LEO), minimising the need for human oversight and enabling rapid response to external changes in an ever more congested space environment.

“LEO orbit will become more commercial in a post-ISS scenario, as there are already consortia developing space stations for this orbit. There will be crewed missions, and they will be commercial. If the system is deployed on Columbus, it can operate as a testing platform for future missions,” says Dr. Temenushka Manthey, Technical Lead for the Demonstrator ORBIT-STAR at Airbus Defence and Space GmbH.

What comes next?

The next step would be deploying an execution platform onboard the real Columbus module, which could be used for a multitude of experiments, and for testing how AI can improve specific subsystems. It can also become the perfect tool for AI pioneers in space, who would be able to conduct blue sky research in a real space environment.

The activity aligns with ESA Vision to improve autonomy, and could be instrumental after its ISS phase in various long term endeavours, including, for example, mitigating communication delays between the Lunar Gateway and Earth ground control.

Discovery programme

Artificial intelligence in space

The activity was funded by the Discovery element of ESA’s Basic Activities.  It sources new ideas for new activities from industry, academia and the general public through the Open Space Innovation Platform (OSIP).

“AI will become integrated into spacecraft if we wish to explore new horizons. Discovery programme offered us resources to explore this topic. It helped us reduce risks when actually building the system, as we were able to look into what is meaningful or not meaningful for further research,” says Temenushka Manthey.



Source link