Powerful new AI model will map the moon's surface in more detail than ever, NASA and IBM scientists say
AstronomyEarth ScienceData Science
THE AI ANGLE
Unifying multimodal planetary data to map lunar topography and resourcesNASA and IBM have developed the open-source Lunar Foundation Model (LFM), an AI trained on decades of disparate lunar mission data compiled into a multi-layered benchmark dataset called SomBench. The model integrates multiresolution optical, elevation, thermal, and gravitational readings alongside explicit solar metadata, outperforming existing vision transformers like SwinV2-B in identifying craters, rare volcanic landforms, and shadowed polar ice deposits. This provides planetary and data scientists with an adaptable, unified backbone to resolve data bottlenecks and support future lunar exploration.
THE TEACHING ANGLE
Instructors can explore how integrating domain-specific physical metadata, like solar angles, into foundation model architectures allows AI to resolve severe environmental distortions, such as airless shadow effects, far better than standard computer vision models.Read the original at livescience.com Generate teaching or study materials
More in Astronomy
- AI’s best coding agent fails 60% of the time — and the data backs it upThe New Stack · September 15, 2026
- Observing Earth from orbit is unpredictable and messy: Could 'liquid' AI clear the view?Phys.org — Technology · September 15, 2026