Covid-19
Advertisement

These 11 Asteroids Might Collide With Earth, Says Neural Network

A new AI neural network has tracked 11 asteroids that can level cities. NASA didn't know.

A team of scientists at Leiden University in the Netherlands developed a neural network — called "Hazardous Object Identifier" — that they claim predicts asteroid collisions with Earth.

RELATED: WHAT IS THE PROBABILITY OF A HUGE CIVILIZATION-ENDING ASTEROID IMPACT?

Neural Network Unto Mass Death

The new AI system selected 11 asteroids not previously identified as hazardous by NASA as dangerous, each larger than 100 meters (328 feet) in diameter — enough mass to transform into an explosive force of hundreds of nuclear weapons, upon impact — which is overkill to any city in the world. Asteroids that could stray within 4.7 million miles of Earth were also sought, according to the paper published in the journal Astronomy & Astrophysics in February.

It could be luck, but none were deemed a threat in the here-and-now. While the likelihood of these asteroids to impact Earth is slim-to-none, they will make flybys sometime between the distant years 2131 and 2923.

The team's supercomputer allowed the researchers to fast-forward through 10,000 simulated years of our solar system — including all its planets. Later, the team reversed the simulation to peek into Earth's future, and see where the space rocks of future apocalypses have been before smashing humanity and destroying our cities.

"If you rewind the clock, you will see the well-known asteroids land again on earth," said Simon Portegies Zwart, astronomer and co-author of the study, in a statement. "This way you can make a library of the orbits of asteroids that landed on earth."

AI with high asteroid impact accuracy

This library of simulated human disasters was also a training ground for the team's neural network, after which the AI extrapolated patterns from the data — to predict bad ends for the asteroid-bound worlds.

The ability of the team's supercomputer to accurately predict real-world threats was tested on a NASA catalog of 2,000 asteroids. Shockingly, it differentiated safe from dangerous asteroids with 90.99% accuracy.

"We now know that our method works, but we would certainly like to delve deeper in the research with a better neural network and with more input," said Zwart. "The tricky part is that small disruptions in the orbit calculations can lead to major changes in the conclusions."

Advertisement

Stay on top of the latest engineering news

Just enter your email and we’ll take care of the rest:

By subscribing, you agree to our Terms of Use and Privacy Policy. You may unsubscribe at any time.

Advertisement