For the past NASA’s Mars Reconnaissance Orbiter has been orbiting the Red Planet for 15 years to study its climate and geology. The orbit sends every day a treasure trove of images and other sensor data that NASA scientists have used to search for safe landing sites for robbers and to understand the distribution of water ice on the planet. Of particular interest to scientists are the crater photos of the orbit, which can provide a window into the deep history of the planet. NASA engineers are still working on a mission to return samples from Mars; without the rocks that will help them calibrate remote satellite data with conditions on the surface, they have to do a lot of advanced guessing when it comes to the age and composition of each crater.
For now, they need other ways to bother the information. One proven method is to extrapolate the age of the oldest craters from the properties of the newest planet. Since scientists can know the age of some recent impact sites within a few years – or even weeks – they can use them as a basis to determine the age and composition of much older craters. The problem is finding them. Searching through image data of a planet in search of the drawing signs of fresh impact is tedious to work with, but it’s exactly the kind of problem an AI was made to solve.
Late last year, NASA researchers used a machine learning algorithm to discover fresh Mars craters for the first time. The AI has discovered dozens of them hiding in image data from the Mars Reconnaissance Orbiter, revealing a promising new way to study planets in our solar system. “From a scientific perspective, it’s exciting because it increases our knowledge of these features,” said Kiri Wagstaff, a computer scientist from NASA’s Jet Propulsion Laboratory and one of the leaders of the research team. “The data was there all the time, it was just that we did not see it ourselves.”
The Mars Reconnaissance Orbiter has three cameras, but Wagstaff and her colleagues trained their AI using images from just the Context and HiRISE footage. Context is a relatively low-resolution gray-scale camera, while HiRISE uses the largest reflective telescope ever sent into deep space to produce images with three times the resolution of Google Maps.
First, the AI conducted nearly 7,000 orbiting photographs of Mars – some with previously discovered craters and others without any – to teach the algorithm how to detect a new strike. After the classifier was able to accurately locate craters in the training set, Wagstaff and her team loaded the algorithm onto a supercomputer at the Jet Propulsion Laboratory and used it to search a database of more than 112,000 images of the orbit.
“There’s nothing new with the underlying machine learning technology,” says Wagstaff. ‘We used a fairly standard convolutional network to analyze the image data, but it’s still a challenge to apply it to scale. That was one of the things we had to contend with here. ”
Most recent craters on Mars are small and may be only a few feet wide, meaning they appear as dark pixel spots on contextual images. If the algorithm compares the image of the candidate crater with a previous photo from the same area and notices that the dark spot is missing, the chances are high that it has found a new crater. The date of the previous image also helps to determine the timeline for when the impact occurred.
After the AI identified promising candidates, NASA researchers were able to make follow-up observations with the high-resolution camera of the orbit to confirm that the craters do exist. Last August, the team received its first confirmation when the orbit photographed a group of craters identified by the algorithm. It was the first time an AI had discovered a crater on another planet. “There was no guarantee that there would be new things,” Wagstaff says. “But there were a lot of them, and one of our big questions is: what makes them harder to find?”