As Seen In Forbes Science | June 17, 2019

Even in this age of high-speed data analysis, a keen human eye normally can’t be beaten when poring over images of potential asteroidal impactors. But Artificial Intelligence (A.I.) could soon change all that. The El Segundo, Calif.-based Aerospace Corporation is now testing A.I. software designed to help astronomers speed up the process of identifying and tracking threatening Near-Earth Objects (NEOs).

NASA’s Planetary Defense Coordination Office already uses numerous telescopes to find and monitor NEOs that might have the potential to impact Earth. But the non-profit Aerospace Corporation’s A.I. team is working with NASA on implementing software dubbed NEO AID (Near-Earth Object Artificial Intelligence Detection) to differentiate false positives from asteroids and comets that might be real threats.

Nightly, researchers at locations such as the Catalina Sky Survey on Mount Lemmon in Tucson, Ariz. pore over hundreds of images of star fields in search of fast-moving objects that need more scrutiny, says Aerospace Corporation. It’s here that Aerospace A.I. engineers used 100 terabytes of data to build and train an artificial intelligence model that is now capable of classifying NEO targets of interest. And by Aerospace Corporation’s calculations, this new A.I. tech has already increased the sky survey’s performance by 10 percent with room for development.

NASA’s Center for Near-Earth Object studies says that with over 90 percent of NEOs larger than one kilometer already discovered, the NEO program is now focusing on finding the 90 percent larger than 140 meters. However, there are still space rocks in the 10- to 20-meter diameter range that comes closer than the distance from the Earth to the Moon. That happens at least once or twice a month and if any of these objects were to actually strike a highly-populated area, they would do significant damage.

But how would this new A.I. tech help NASA in its current search?

NASA still relies on human eyes to determine a NEO’s threat assessment. The hope is that A.I. can streamline that process by classifying an image as high priority or low priority.

As an example of input data, Aerospace scientists analyze thousands of images of candidate Near Earth Objects in sequences of four 30 second exposures.

CREDIT: UNIVERSITY OF ARIZONA, CATALINA SKY SURVEY

The software technology that Aerospace Corporation has developed reduces the number of false identifications that human observers have to review, Jon Neff, Aerospace Corporation’s senior project leader for artificial intelligence, analytics, and innovation department, told me. Human eyes and brains are very good at finding small differences in images, he says.

By training neural networks to imitate the way humans classify images of the night sky taken by telescopes, astronomers can automatically identify objects with a high probability of being NEOs. But Neff is quick to point out that this new NEO AID technology is designed to complement current methods of measurements, not replace them.

As for the tech’s costs?

The prototype software cost about $50,000 to develop, says Neff. We think an operational software system would cost about $500,000 , he says.

How would this tech help in preventing the kind of high-inclination civilization-ending comets to which astronomers are still mostly blind?

As with cancer, the key to survival is early detection and diagnosis, says Neff. The first step is to point telescopes where they can find high inclination objects; the next step is sorting through millions of images to find the small number of objects that pose a threat, he says.

“Our technology can help save civilization by increasing the rate at which new objects are detected,” said Neff. “If we can detect threats early enough, we may have time to deflect them.”

Follow Bruce Dorminey on Facebook, Twitter and Google +. And like his ‘Distant Wanderers’ exoplanet Facebook page.