Planning a space mission takes years of scientific preparation and data analysis. As space agencies race to set up a lasting presence on the moon, researchers at The Ohio State University say artificial intelligence could make this work faster and easier. A new AI framework called LUMINA helps scientists judge possible lunar landing sites, spot hazards future crews might face, and search new regions for useful geologic features, according to an Ohio State News release.
LUMINA is built on an earlier lunar vision-language foundation model, an AI trained to link images of the moon with language so it can describe terrain and answer scientific questions. LUMINA goes further with an agentic framework that plans and carries out several analysis steps on its own, then explain the evidence behind its recommendations. Preliminary results show a strong ability to make complex lunar observations and generate useful commentary for planners, which could help locate areas rich in water ice and other resources.
An assistant, not a replacement
The system is meant as decision support, not a machine that makes final calls. Most AI tools are passive, she notes: you ask a question and get an answer. The goal is an active scientific assistant that works through problems alongside human experts, who still make the final judgments.
The approach could reach beyond the moon. New models would generalizesto other surfaces, such as Mars, some asteroids, and even Earth. As datasets grow, planners will need efficient tools to search them; otherwise finding the best places to land is like looking for a needle in a haystack.
Next, the team plans to test LUMINA on new high-quality lunar data and check how well it supports mission planning. NASA's Artemis program creates important opportunities, because fresh lunar data can improve systems like this one.
This matters because future missions will produce millions of images and datasets, more than people can review by hand. An AI assistant that can sift through this information, flag safe and resource-rich landing sites, and explain its reasoning could save time and lower risk for crews. If it generalizes, the same kind of tool could aid exploration across the solar system.