IBM launches open-source AI model for NASA lunar research

IBM and NASA have launched the NASA-IBM Lunar Foundation Model, an open-source AI system available on Hugging Face. It is designed to help researchers analyze petabytes of lunar observation data and identify geological features relevant to future exploration.

Trained on a multimodal NASA dataset, the model can be adapted to study different lunar features instead of requiring a separate model for each task. IBM and NASA say it has identified craters and volcanic features more accurately and reduced errors in locating potential ice deposits.

The dataset will be released alongside the model. It is described as the first unified, publicly available cache prepared for machine learning, combining over 30 spatially aligned layers from nine instruments across four missions. It includes tens of thousands of images and maps from NASA’s Lunar Reconnaissance Orbiter (LRO) and GRAIL mission.

IBM, which has worked with NASA for over five decades, including on the Apollo missions, says the model could help researchers study lunar geology, assess possible landing locations and support future efforts to identify resources such as ice deposits. NASA-IBM Lunar Foundation Model capabilities may also help scientists examine volcanic history, thermal evolution and the Moon’s geological past.