Google-backed AI successfully detects two hidden planets bigger than Earth

27 Mar 2019

Image: © dottedyeti/Stock.adobe.com

AI has once again proven its worthiness as a planet hunter with the discovery of two new worlds that have previously evaded detection.

Given the vastness of the universe, it isn’t at all surprising that sometimes we might miss a planet or two in our search for Earth-like worlds. However, as a new breakthrough has shown, artificial intelligence (AI) can help us find these worlds.

In a paper soon to be published to The Astronomical Journal, astronomers from The University of Texas at Austin – in partnership with Google – announced the discovery of two hidden planets within the vast archive of NASA’s Kepler space telescope’s K2 mission. Until now, previous efforts – both human and computer-aided – had simply missed them during scans, but a new algorithm designed by the astronomers is able to see what many cannot.

This totally new algorithm was needed for the search, according to its designer Anne Dattilo, because data taken during Kepler’s extended K2 mission differs significantly from that collected during the spacecraft’s original mission.

After a mechanical failure on board the spacecraft, data collected during the K2 mission was significantly altered because Kepler was changing position frequently. While mission planners found a workaround, the spacecraft was left with a wobble that AI had to take into account.

‘There are a lot of planets out there that we don’t see’

The new planets located in the Aquarius constellation discovered by the AI included K2-293b, which is a planet orbiting a star 1,300 light years away, and K2-294b, a planet orbiting a star 1,230 light years away.

Dattilo said they are both very typical of planets found in K2. “They’re really close in to their host star, they have short orbital periods and they’re hot,” she said. “They are slightly larger than Earth.”

Explaining how important this algorithm and other AI will be in the search for Earth-like planets, astronomer Andrew Vanderburg said: “Even if every star had an Earth-sized planet around it, when we look with Kepler we won’t find all of them.

“That’s just because some of the data’s too noisy, or sometimes the planets are just not aligned right. So, we have to correct for the ones we missed. We know there are a lot of planets out there that we don’t see for those reasons.”

Looking to the future, Dattilo said her current algorithm can search through the entire K2 dataset, equating to approximately 300,000 stars. Also, it should be easily applicable to Kepler’s successor, TESS, launched less than a year ago.

Colm Gorey was a senior journalist with Silicon Republic

editorial@siliconrepublic.com