A research team at Julius Maximilians University Würzburg (JMU) is developing an AI system with the NOVA project (FPGA-based resilient earth observation and reconnaissance) that will analyze satellite images directly in orbit in the future. The aim is to transmit only relevant information back to Earth, thereby using the limited communication capacities of satellites more efficiently.

AI decides which data to transmit

Instead of sending all image data to ground stations, the system processes the images directly onboard the satellite. The integrated Artificial Intelligence recognizes, for instance, forest fires, ships, or other relevant objects and transmits only important information like coordinates or selected image excerpts. This should allow critical events to be detected much more quickly and relieve communication links.

Adaptable AI for changing missions

A key feature of the project is the flexible adaptation of the AI during the satellite mission. The image processing is based on a Field-Programmable Gate Array (FPGA) – a programmable special chip that can also be reconfigured in orbit. This way, AI models and image analysis procedures can be adapted to new tasks during the mission, such as for environmental monitoring, disaster response, or scientific earth observation.

Use in environmental and security applications

The payload computer responsible for image processing is being developed by the Munich space company Engineering Minds Munich. The project partners aim to create a powerful and energy-efficient platform that will make satellites more flexible and autonomous in the future. Potential applications range from early detection of forest fires to environmental and climate monitoring and reconnaissance in crisis situations.

The NOVA project is funded by the German Aerospace Center (DLR) with funds from the Federal Ministry for Economic Affairs and Energy (BMWE).