Home / Science / AI for the Moon: NASA & IBM Launch Lunar Foundation Model
AI for the Moon: NASA & IBM Launch Lunar Foundation Model
10 Sep
Summary
- A new AI model aids moon exploration under NASA's Artemis program.
- It analyzes lunar data, outperforming older methods by up to 23%.
- The model and a lunar dataset are now publicly available.
NASA and IBM have jointly released the Lunar Foundation Model, a pioneering AI tool aimed at advancing moon exploration under NASA's Artemis program. This model is now accessible on the open-source platform Hugging Face.
It was trained using extensive lunar observation data, identified by IBM and NASA researchers. The AI's purpose is to provide insights that will enhance human presence on the moon.
Previously, scientists manually sifted through vast amounts of lunar data or relied on limited, task-specific machine learning models. The new foundation model significantly streamlines this process by identifying patterns across various data sources at scale.
This advanced AI can identify critical lunar geographic features, such as craters, volcanic formations, and potential ice deposits, with an accuracy improvement of up to 23% compared to conventional methods. These findings are vital for planning future Artemis missions.
In addition to the AI model, IBM and NASA have also introduced the first open-source lunar dataset. This dataset comprises tens of thousands of maps and images gathered from nine instruments across four different lunar missions, fostering broad scientific access and collaboration.
