NASA and IBM open source lunar mapping tools

NASA and IBM have joined forces to release an open-source AI model of the Moon, called the NASA-IBM Lunar Foundation Model. The model has been trained on a vast dataset of lunar observations, curated by researchers from both organizations. This AI model integrates data from various sources, including different observational modalities, angles, and spatial scales.

Scientists can now use the model to analyze lunar features more efficiently and discover previously unidentified lunar ice deposits, volcanic features, and craters. The presence of lunar ice suggests the potential for water and oxygen, which could be beneficial for future manned missions. The model was built using a lunar dataset containing over 30 spatially-aligned layers, generated from data collected by nine instruments in four missions.

NASA's chief science data officer, Kevin Murphy, emphasized the need to make data more accessible for exploration and use. IBM's director of research for Europe, Juan Bernabe-Moreno, highlighted that the model provides a foundation for scientists to explore the Moon at scale, revealing patterns difficult to see in isolation. This initiative follows previous collaborations between NASA and IBM, such as the release of Prithvi, an open-source foundation AI model for analyzing satellite imagery, and an AI climate model.

The Lunar Foundation Model is available for download, but the specific hardware requirements depend on the application, the size and number of inputs, and whether the model is being trained or used for predictions.