Ohio State Researchers Develop Lunar Landing AI Model

The new LUMINA framework evaluates lunar terrain to identify optimal sites for future human exploration missions.

Updated on Oct. 31, 2026 in Artificial Intelligence

Isometric editorial illustration of simplified lunar crater topography featuring geometric elevation markers, representing AI-assisted planetary site analysis. — Ohio State researchers have introduced LUMINA, an agentic AI framework designed to analyze lunar topography and identify optimal landing sites for future exploration. AI Illustration. Upload story photo >

Live Poll

Do you support using artificial intelligence to plan and manage future human space missions?

Researchers at The Ohio State University introduced LUMINA, an agentic AI framework designed to assess potential lunar landing sites. The model identifies locations by analyzing topography, terrain hazards, and water ice availability to support mission planning.

Why it matters

As NASA prepares for future missions, planners require sophisticated tools to evaluate surface data and ensure the safety of human habitation outposts. This technology aims to streamline the selection of landing sites that maximize scientific value and operational effectiveness.

Built upon the Large Language-and-Vision Assistant for Lunar Exploration, or LLaVA-LE, the model incorporates orbital imagery and surface data to map hazards. The underlying LLaVA-LE model is currently available for public access via GitHub.

The players

The Ohio State University

This public research university serves as the institution where the researchers developed the LUMINA artificial intelligence framework.

NASA

The United States government agency is responsible for the national space program and the upcoming 2028 mission to return humans to the moon.

The details

The LUMINA model processes complex geological data to determine sites that provide necessary resources like water ice. Beyond lunar applications, researchers expect the agentic framework to assist in the development of future outposts on other planetary bodies, including Mars.

1972: The year of the final Apollo 17 lunar mission.
June 2026: The LLaVA-LE model is presented at the CVPR2026 AI4Space Workshop.
October 2026: The LUMINA study is presented at the 58th Annual Meeting of the Division for Planetary Sciences.
2028: The year NASA plans to return astronauts to the lunar surface.

The Tech Race

The transition to agentic AI frameworks represents a shift from static satellite analysis to automated, real-time site evaluation for space agencies. This technology positions researchers to move beyond the limitations of legacy mapping tools used during the 1970s.

While primarily focused on space mission planning, the framework provides a new open-source tool for developers interested in applying vision-based AI to complex terrain mapping. Future adaptations of this technology could eventually influence how remote sensing data is processed for terrestrial mapping applications.

The takeaway

Advancements in agentic AI models like LUMINA are becoming essential for managing the high-stakes logistics of extraterrestrial mission planning. Researchers can leverage these open-source tools to improve decision-making accuracy when dealing with hazardous planetary environments.