NASA and IBM have released an open-source artificial intelligence (AI) model named the ‘NASA-IBM Lunar Foundation Model’ designed to aid lunar observation. This model will assist researchers in identifying potential ice deposits in the Moon’s permanently shadowed regions, locating safe landing sites, and studying volcanic structures. According to NASA and IBM, testing has shown that the model identifies key features on the lunar surface 23% faster than other methodologies. Kathmandu, 11 September.
The American space agency NASA launched this new open-source AI model to analyze decades of lunar observation data and strengthen plans for sustained human presence on the Moon. Developed through collaboration between NASA and IBM, the ‘NASA-IBM Lunar Foundation Model’ is publicly available and trained on over 30 data layers collected by nine instruments from previous NASA missions. The dataset also includes information gathered by the Lunar Reconnaissance Orbiter (LRO).
This model is comparable to IBM and NASA’s ‘Earth’ series of open foundation models, which have been applied successfully across geospatial and meteorological domains. The new model enables researchers to pinpoint potential ice locations in the Moon’s permanently shadowed areas, map craters suitable for safe landings, and analyze volcanic features. Traditionally, such tasks required scientists to manually study maps and images or rely on lower-resolution machine learning tools.
NASA and IBM report that during testing, this model identified major lunar surface features up to 23% faster and with greater accuracy compared to existing methods. The presence of ice on the Moon indicates potential sources of water and oxygen, which are of significant interest to space agencies. These resources will be critical for constructing permanent bases or stations on the lunar surface and for producing rocket fuel necessary for future Mars missions. Under NASA’s Artemis program, astronauts are planned to return to the Moon during 2026-2027, fostering the development of new technologies for long-term lunar presence and upcoming Martian exploration. This AI model is freely accessible on the Hugging Face platform, with its code also hosted on GitHub, and is expected to support future lunar research efforts.
