Researchers from the University of Adelaide and Australia's SmartSat CRC have successfully launched a geospatial AI model into space for the first time in history. The Prithvi Geospatial model, created by NASA and IBM, was tested using the Kanyini satellite and the IMAGIN-e module on the International Space Station (ISS). This system demonstrated the ability to detect clouds and floods directly in space without sending data back to Earth. This is reported by Ixbt.com reports.
The Prithvi model was trained on 13 years of satellite observations, including data from NASA's Landsat and the European Space Agency's Sentinel-2. Thanks to its open-source architecture, this foundation model is adapted for a wide range of tasks, from monitoring natural disasters to assessing agricultural yields.
Experts note that with traditional methods, satellites have to transmit massive amounts of data to Earth, which overloads communication channels. Foundation models like Prithvi allow for analysis directly on the satellite, drastically reducing data transmission requirements and increasing system flexibility.
NASA representative Kevin Murphy emphasized that the agency's main goal in providing its AI models as open source is to accelerate scientific progress. In the future, operators will be able to communicate with satellites and ask them directly about system status or observation data.
NASA plans to further develop this technology. Having introduced the Surya model for solar research in 2025, the agency is working on creating similar intelligent systems for astrophysics, planetary science, and biology in the near future.
Read “Zamin” on Telegram!
