In late last year, NASA announced that it would train artificial intelligence (AI) to help find fresh craters on Mars. According to a report from Science Times a few months ago, planetary scientists and AI researchers are collaborating on making a machine learning tool that will reduce lead time and increased the findings on the surface of Mars.

NASA's Mars Reconnaissance Orbiter has been doing laps around the Red Planet for 15 years to study the planet's climate and geology. It sends back photos of Mars every day which is used by NASA scientists to scout safe landing sites for the rovers.

But they are most interested in the planet's craters because they believe that studying about them will tell the deep history of Mars. They rely on missions to return samples from Mars to determine the crater's age and composition.

AI Crater Classifier Discovered Dozens of Fresh Craters on Mars

The AI crater classifier used by NASA was fed nearly 7,000 photos from the Mars Reconnaissance Orbiter to teach the algorithm how to detect a fresh strike.

The orbiter carries three cameras but the AI was trained using images from just the Context, a relatively low-resolution grayscale camera, and HiRISE imagers that use the largest reflecting telescope ever sent to space to produce images three times higher the resolution of images used on Google Maps.

When the AI could accurately detect craters in the training set, NASA's Jet Propulsion Laboratory computer scientist Kiri Wagstaff and her team loaded the algorithm onto a supercomputer at JPL and used it to carefully search craters from more than 112,000 images from the orbiter.

Recently, NASA revealed that the AI discovered dozens of fresh craters hiding in the image from the orbiter and said that they have a new promising way to study planets throughout the solar system.

According to a Wired report, the fresh craters discovered were only a few feet across that it appears as dark pixelated blotched on Context images. But comparing it with an earlier photo from the same area and finds that there is no dark spot there, the scientists believe that the AI successfully found a new crater.

"From a science perspective, that's exciting because it's increasing our knowledge of those features," Wagstaff said. "The data was there all the time, it's just that we hadn't seen it ourselves."

RELATED ARTICLE: A Machine Learning Tool Supports the Search for New Craters on Mars


Why Is It Important to Study the Craters on Mars?

For the last 15 years, NASA scientists had to scour the images from the orbiter for about three-quarters of an hour for a single image. But with the AI crater classifier, that task is now easier to accomplish.

Planetary scientist Ingrid Daubar from Brown University, who collaborated on this project, said that studying the craters of Mars does not only help determine the age of the Martian surface but also teach scientists about what is beneath it. By studying the exposed water ice and how it disappeared over time, scientists were able to come up with ideas about how the ice was distributed throughout the planet.

Daubar hopes that the AI that regularly scours images for fresh craters could alert scientists within days or weeks of their formation that will teach them more about the history of Mars.

READ MORE: [LOOK] Images From Mars Reveal Unusual Triple Craters Overlapping Each Other


Check out more news and information on Mars in Science Times.