NASA and IBM unveil open-source lunar AI model

The AI model could help scientists better identify lunar ice, volcanos and craters.

A new open-source AI model trained on NASA’s extensive lunar observation data is expected to help scientists better navigate moon colonisation plans.

The NASA‑IBM Lunar Foundation Model is one of the first publicly available foundation models for the scientific exploration of the moon, coming just months after the two organisations’ Prithvi AI made history by becoming the first geospatial foundation model to be deployed in orbit.

The Earth’s lunar neighbour has an arid and varied topography covered in craters, ice and volcanic activity – all picked up over the decades by various sensors and instruments. The petabytes of collected data, is, however, scattered across maps and images that need physical examination, or are accessible only by task-specific machine learning models.

A domain-specific AI model could change how scientists analyse data, potentially accelerating how lunar geography is identified and understood. The Lunar Foundation Model achieves this by combining multimodal and multi-resolution observations.

According to IBM and NASA, the new model can be used to investigate potential lunar ice deposits in the moon’s permanently shadowed regions, which could indicate the presence of water and oxygen.

It could also be used to study lunar volcanic features, called ‘irregular mare patches’, and better identify craters to ensure safer landing sites for spacecraft.

“NASA has spent decades building an extraordinary scientific record of the moon, but collecting data is only part of the job,” said Kevin Murphy, the chief science data officer and acting chief data and AI officer at NASA.

“We also have to make data easier for scientists to explore and use. The NASA-IBM Lunar Foundation Model shows what’s possible when we bring AI to NASA’s petabytes of scientific data. That’s a real opportunity we see with AI – turning large-scale data into new discoveries.”

Alongside the model, IBM and NASA also scientists built the first open-source lunar dataset of its kind, made out of layers of data from nine instruments used across four missions.

The dataset includes tens of thousands of images and lunar surface maps from NASA’s Lunar Reconnaissance Orbiter and the GRAIL mission, and data from the Japanese Aerospace Exploration Agency.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.