For the past For 15 years, NASA’s Mars Reconnaissance Orbiter orbits the red planet studying its climate and geology. Each day, the orbiter sends a treasure trove of images and other sensor data that NASA scientists have used to search for safe landing sites for rovers and understand the distribution of water ice on the planet. Photographs of the orbit crater are of special interest to scientists, who can provide a window into the deep history of the planet. NASA engineers are still working on the mission to return samples from Mars; without the rocks that will help them calibrate remote satellite data with surface conditions, they will have to make many conjectures when determining the age and composition of each crater.
For now, they need other ways to provoke this information. A proven and true method is to extrapolate the age of the oldest craters from the characteristics of the newest on the planet. Since scientists can know the age of some recent impact sites in a few years — or even weeks — they can use them as a basis for determining the age and composition of much older craters. The problem is finding them. Counting through the value of a planet with image data looking for telltale signs of a new impact is tedious work, but it is exactly the kind of problem that was done to solve an AI.
Late last year, NASA researchers used a machine learning algorithm to discover fresh Martian craters for the first time. The AI discovered dozens of them hidden in imaging data from the Mars Reconnaissance Orbiter and revealed a promising new way to study the planets in our solar system. “From a scientific perspective, it’s exciting because it increases our knowledge of these characteristics,” says Kiri Wagstaff, a computer scientist at NASA’s Jet Propulsion Laboratory and one of the leaders in the research team. “The data was there all the time, it’s just that we hadn’t seen it.”
The Mars Reconnaissance Orbiter carries three cameras, but Wagstaff and colleagues trained their artificial intelligence using only Context and HiRISE images. Context is a relatively low-resolution grayscale camera, while HiRISE uses the largest reflecting telescope ever sent into deep space to produce images with resolutions approximately three times higher than the images used in Google Maps. .
First, the AI was fed nearly 7,000 orbiting photos of Mars, some with previously discovered craters and some without any, to teach the algorithm how to detect a new attack. After the classifier could accurately detect the craters in the training set, Wagstaff and his team loaded the algorithm into a supercomputer at the Jet Propulsion Laboratory and used it to comb through a database of more than 112,000 images of the orbiter.
“There’s nothing new with the underlying machine learning technology,” Wagstaff says. “We have used a fairly standard convolution network to analyze the image data, but being able to apply it to scale is still a challenge. That was one of the things we had to fight for here. “
Most recent craters on Mars are small and can be only a few meters in diameter, meaning they appear as dark pixelated spots in context images. If the algorithm compares the image of the candidate crater with a previous photo of the same area and finds that it lacks the dark patch, there is a good chance that it will find a new crater. The date in the image above also helps establish the chronology of when the impact occurred.
Once the AI identified some promising candidates, NASA researchers were able to make some follow-up observations with the orbit’s high-resolution camera to confirm that the craters actually existed. Last August, the team obtained its first confirmation when the orbit photographed a cluster 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 many, and one of our big questions is, what makes them harder to find?”