NASA engineer teaches AI to be 'GPS-like' to guide astronauts over the lunar surface

An AI system is being developed to lead explorers around the lunar surface. 
Sade Agard
LunaNet concept graphic
LunaNet concept graphic

NASA/Reese Patillo 

Without instruments like the GPS we have on Earth, scientists have been attempting for years to figure out how to travel over the lunar surface.

Since the Moon's atmosphere is significantly thinner than Earth's, it is challenging to determine the size and distance of distant landmarks when looking at the horizon. 

Landmarks like trees or buildings on Earth can serve as fuzzy but useful distance measures- features that are non-existent on the Moon. 

Teaching AI to be 'GPS-like'

An astronaut on the Moon would also have trouble navigating the vast, unexplored area because there is no atmosphere to scatter light. Thanks to the Sun's powerful beams, these cause vision and depth perception to be altered.

Now, Alvin Yew, a research engineer at NASA's Goddard Space Flight Center in Greenbelt, Maryland, is creating an AI system that leads explorers around the lunar surface. 

Like how our GPS pinpoints places on Earth, Yew is teaching artificial intelligence to mimic lunar horizon features as they would appear to a lunar explorer.

The technology was developed using data from NASA's Lunar Reconnaissance Orbiter. More specifically, it utilizes the Lunar Orbiter Laser Altimeter (LOLA), which measures slopes and lunar surface roughness. Simply put, LOLA generates high-resolution topographic maps of the Moon. 

These digitized panoramas are then exploited to match images captured by a rover or astronaut with known boulders, ridges, and even craters, allowing precise location identification for any given area.

Most Popular

LunaNet: Complementing the Moon's upcoming internet

"For safety and science geotagging, it's important for explorers to know exactly where they are as they explore the lunar landscape," said Alvin Yew in a NASA press release

"Conceptually, it's like going outside and trying to figure out where you are by surveying the horizon and surrounding landmarks," Yew said. 

"While a ballpark location estimate might be easy for a person, we want to demonstrate accuracy on the ground down to less than 30 feet (9 meters). This accuracy opens the door to a broad range of mission concepts for future exploration."

Yew's AI system will also support LunaNet, the Moon's upcoming internet. According to NASA researchers, the collection of lunar satellites seeks to provide internet connectivity comparable to that on Earth. 

The idea is that LunaNet will serve as a network that spacecraft and future astronauts can access without organizing data transfers in advance, as space missions do presently. 

A 'back-up' for dangerous space missions

Still, Yew's AI system would not so much help out in the sense of providing internet. Instead, it could serve as a geolocation backup should communication signals not be available, particularly in dangerous space missions.  

According to NASA, Yew's system could be developed into a handheld device that would combine AI and GIANT's capabilities (Goddard Image Analysis and Navigation Tool). 

GIANT is an optical navigation tool that successfully verified navigation data for NASA's OSIRIS-REx mission. This went on to collect a sample from asteroid Bennu.

Yew told PopSci that while the AI is still being developed, he is already making improvements. The system will first be put through its paces in a virtual setting before (hopefully) using actual lunar landscape data from one of the Artemis missions.

"We want to test the robustness of the algorithm," he added. Yew described this as throwing a person (or robot) anywhere on the Moon and being able to locate them anywhere. Imagining that little yellow man on Google Street Maps much?