Trees in front of moon

AI ‘eye’ could help explore features on the Moon

Image credit: Catiamadio/Dreamstime

A Moon-scanning method that can automatically classify important lunar features from telescope images could significantly improve the efficiency of selecting sites for exploration.

The choice of future landing and exploration sites on the Moon may come down to the most promising prospective locations for construction, minerals, or potential energy resources. But scanning across a large area, looking for features a few hundred metres across, by eye, is laborious and often inaccurate, experts have said, which makes it difficult to pick optimal areas for exploration.

Siyuan Chen, Xin Gao, and Shuyu Sun at King Abdullah University of Science and Technology (KAUST) in Saudi Arabia, along with colleagues from the Chinese University of Hong Kong, have now applied machine learning and artificial intelligence (AI) to automate the identification of prospective lunar landing and exploration areas.

“We are looking for lunar features like craters and rilles, which are thought to be hotspots for energy resources like uranium and helium-3 – a promising resource for nuclear fusion,” Chen explained. “Both have been detected in Moon craters and could be useful resources for replenishing spacecraft fuel.”

Although machine learning is a very effective technique for training an AI model to look for certain features on its own, the first problem faced by Chen and his colleagues was that there was no labelled dataset for rilles to help train their model. But Chen said they overcame this by constructing their own dataset with annotations for both craters and rilles.

“To do this, we used an approach called transfer learning to pre-train our rille model on a surface crack dataset with some fine-tuning using actual rille masks,” Chen said. “Previous approaches require manual annotation for at least part of the input images;  our approach does not require human intervention and so allowed us to construct a large, high-quality dataset.”

The researchers then developed a computational approach that could identify both craters and rilles at the same time, something that had not been done before, according to the research team.

“This is a pixel-to-pixel problem for which we need to mask the craters and rilles in a lunar image,” Chen explained. “We solved this problem by constructing a deep-learning framework called high-resolution-moon-net, which has two independent networks that share the same network architecture to identify craters and rilles simultaneously.”

According to the researchers, their approach achieved precision as high as 83.7 per cent, higher than existing state-of-the-art methods for crater detection.

Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.

Recent articles