Artificial intelligence continues to amaze with its capabilities.
Over the course of several billion years, thousands of meteorites have been falling on the moon’s surface, leaving behind large craters. At the moment, scientists do not know their exact number, but they are working to fill this knowledge gap.
Recently, Chinese scientists studied data collected by spacecraft and used artificial intelligence to count the number of craters in numerous images. In total, more than 109,000 craters were discovered, which differ from each other not only in shape but also in age.
Based on the data obtained, scientists made very interesting conclusions, which we will discuss below. Researchers also have big plans to improve the work of artificial intelligence in the future. But let’s talk about everything in order.
How did artificial intelligence discover more than 100,000 new craters on the Moon?
As a start, the researchers used the largest database on the moon to count the craters. Numerous photographs of the lunar surface have been taken as part of China’s space missions.
A team of scientists led by Chen Yang first trained a neural network to find craters in photographs, where the grooves were marked in advance. After that, the computer knew roughly how exactly the craters looked in photographs. Accordingly, artificial intelligence was able to cope with the search for craters in completely new photographs.
Counting craters on the moon by hand would take a huge amount of time. In addition, the depressions on the lunar surface are very different and we simply cannot grasp all the features of the image and quickly determine whether it is a crater or not.
But the computer coped with this task quickly and with high accuracy. In total, it managed to recognize 109,956 new craters on the Moon.
The researchers reported that, for the most part, lunar craters are small in size. But by earthly standards, these are real giants, because some of these “small” ones are from 1 to 100 kilometers in diameter.
Artificial intelligence also found several 550-kilometer craters, but initially, they were clearly smaller. The fact is that they have an irregular shape and clearly increased due to landslides and other processes.
Based on the size and depth, scientists were able to determine the age of some of the craters. Among them were those who formed about 4 billion years ago. That is, during the formation of the Earth, craters were already formed on the Moon. However, in those distant times, the fall of meteorites was a common occurrence on Earth too.
The Future of Artificial Intelligence in Space Exploration
There are probably many more craters on the moon than currently thought. Recently, the Chinese apparatus “Chang’e 5” collected not only the lunar soil but also other data on the lunar surface.
Scientists want to use this information for further studies with artificial intelligence. It is expected that this will increase its accuracy and in the already studied photographs, the computer will be able to find even more craters. It can also be used to count craters on other planets like Mars.
By studying the shape and age of the craters, experts can learn more about the development of the solar system. At a minimum, they will be able to figure out which meteorites fell on the moon and other planets and what consequences this led to.
Also, do not forget that in the future, people want to build bases on the Moon and Mars. And before flying to distant objects, it is important to know what there is and what could potentially endanger such colonies. The information obtained will be useful when choosing a place for planting and building structures.
Join the discussion and participate in awesome giveaways in our mobile Telegram group. Join Curiosmos on Telegram Today. t.me/Curiosmos
• Pester, P. (2020, December 22). Moon has way (way) more craters than we thought.
• Researchers identify over 109,000 impact craters on moon. (n.d.).
• Yang, C., Zhao, H., Bruzzone, L., Benediktsson, J., Liang, Y., Liu, B., . . . Ouyang, Z. (2020, December 22). Lunar impact crater identification and age estimation with Chang’E data by deep and transfer learning.