IBM and NASA have released a new open-source lunar foundation model designed to help researchers analyze decades of geographical and environmental data from the Moon, according to announcements from both organizations. Available on Hugging Face, the AI model allows scientists to process petabytes of information gathered over the past five decades, supporting ongoing efforts to establish a sustained human presence on the lunar surface.
Training Data and Public Dataset Release
The NASA-IBM Lunar Foundation model was trained on a multimodal dataset that has also been made publicly available for machine learning research. According to IBM Research, the dataset brings together more than 30 spatially aligned layers from nine instruments across four missions, incorporating tens of thousands of maps and images from NASA’s Lunar Reconnaissance Orbiter (LRO) and NASA’s GRAIL mission. Juan Bernabe-Moreno, Director of IBM Research Europe, UK and Ireland, stated that uncovering the mysteries of the Moon requires an ability to learn from an extraordinary volume of scientific data, adding that the platform gives scientists a foundation to explore the Moon at scale by connecting observations across multiple instruments.
Applications in Lunar Research and Navigation
Researchers typically rely on manual analysis or smaller, task-specific AI models that can be computationally demanding and lack high-precision accuracy. According to TechRadar reporting, the new foundation model allows scientists to adapt a single system to investigate various geologic features rather than building separate models for individual tasks. The model has demonstrated improved accuracy in identifying volcanic features, locating potential ice deposits, and mapping craters. Identifying ice deposits is critical for securing water and oxygen resources for future astronauts, while analyzing volcanic regions and craters helps researchers assess thermal evolution, surface composition, and safe landing sites free of steep slopes or boulders.
Historical Context and Collaboration
The release builds on a five-decade partnership between IBM and NASA that includes collaboration on the Apollo missions. By providing a unified, machine-learning-ready cache of lunar data, the open-source initiative aims to accelerate global research efforts as space agencies prepare for future crewed missions to the Moon and beyond.
Related reading