NASA and IBM's Prithvi AI Model Deployed in Orbit for First Time
Similar Articles
New Tool Earth Index Opens Satellite Analysis to the Public
New AI Technique Maps Ocean Currents in Near Real-Time Using Existing Satellites
Astronomers Confirm Over 100 Exoplanets Using New AI System
NASA and Partners Deliver Next-Generation Spaceflight Computer for Future Missions
India's Skyroot Aerospace Prepares for First Orbital Launch After Reaching $1.1 Billion Valuation
A research team has successfully uploaded and tested the NASA and IBM open-source Prithvi geospatial AI foundation model on two orbiting platforms. This marks the first deployment of a geospatial foundation model in space, where it demonstrated capabilities for tasks like flood and cloud detection. The model, trained on 13 years of global satellite data, could be updated in orbit with minimal bandwidth.
Facts First
- Prithvi is the first geospatial foundation model deployed in orbit, tested on the Kanyini satellite and an ISS payload.
- The model was trained on 13 years of global data from NASA's Landsat and ESA's Sentinel-2 satellites.
- In-orbit testing validated performance for flood and cloud detection across different platforms and computing environments.
- Prithvi can be adapted for mapping flood plains, monitoring disasters, and predicting crop yields.
- Foundation models can be updated in orbit by uploading a small decoder package, conserving bandwidth.
What Happened
A team from Adelaide University and the SmartSat Cooperative Research Center (CRC) uploaded and demonstrated the NASA and IBM open-source Prithvi geospatial artificial intelligence (AI) foundation model aboard two in-orbit platforms. The model was deployed to the South Australian government’s Kanyini satellite and to the Thales Alenia Space IMAGIN-e payload on the International Space Station (ISS). The deployment tested Prithvi's performance in flood and cloud detection across the different orbiting platforms and computing environments. Prithvi was developed by data scientists from IBM and NASA’s IMPACT team and is funded by the Office of the Chief Science Data Officer within NASA’s Science Mission Directorate.
Why this Matters to You
This advancement could lead to faster and more detailed information about environmental changes and disasters. The model's ability to monitor floods and predict crop yields from orbit may improve early warning systems and agricultural planning, potentially helping to safeguard communities and stabilize food supplies. The technology's efficiency means updates can be sent to the orbiting AI with less bandwidth, which could make such systems more responsive and cost-effective to operate.
What's Next
The successful test paves the way for further refinement and application of the Prithvi model. Researchers may now work to adapt the model for specific, real-time monitoring tasks like tracking wildfires or assessing drought conditions. The demonstrated update mechanism suggests future models could be improved or specialized in orbit without needing to be fully replaced, which is likely to be a focus for ongoing development.