How can Machine Learning aid the search for UFOs

In 2017, we caught our first glimpse of something from beyond our solar system, this thing called 1I/’ Oumuamua. It zipped past Earth on its way out, and folks started guessing wildly about what it might be. Based on the little info we had, it didn’t look like anything astronomers had seen before. Some folks even threw out the wild idea that it could’ve been an alien probe or maybe just a broken spaceship bit. The excitement about “alien visitors” got another boost in 2021 when the ODNI released its UFO Report.

So, what happened was they made looking into Unidentified Aerial Phenomena (UAP) a real scientific thing instead of some secret government deal. Now, scientists are gazing at the sky and stuff in space, trying to figure out how they can use fancy computer stuff, AI, and tools to spot possible “visitors.” Like, the folks from the University of Strathclyde had this rad idea – they said, “Let’s use hyperspectral imaging and machine learning together to build a super smart system for figuring out UAP.”

Also Read: How astronomers plan to change an asteroid into a space habitat in just 12 years

Using Machine Learning to classify UFOs

Yep, the team was led by Massimiliano Vasile, a real expert in mechanical and aerospace engineering. They had folks from both the Mechanical and Aerospace Engineering and Electronic and Electrical Engineering departments at the University of Strathclyde, and they teamed up with the Fraunhofer Centre for Applied Photonics in Glasgow. They put together a paper titled “Space Object Identification and Classification from Hyperspectral Material Analysis,” and it’s already up on the internet. Now, they’re just crossing their fingers, hoping Nature Scientific Reports will give it the green light for publication.

This latest study is just one in a series of papers exploring the use of hyperspectral imaging for space-related things. The first paper, titled “Smart Identification of Space Objects with Hyperspectral Imaging,” was published in Acta Astronautica in February 2023. It was a part of the HyperSST project, which wanted to use hyperspectral imaging to track space debris. The UK Space Agency loved it so much that they provided funding, and it paved the way for the European Space Agency’s HyperClass project, which focused on using hyperspectral technology to classify space debris.

In their newest paper, they looked into how they could apply this imaging trick to the whole UFO identification business. Basically, it’s all about gathering and crunching data from the full range of light, usually just to figure out what’s in a picture. Vasile told Universe Today that hyperspectral imaging combined with machine learning could help sort out the real UFO stuff from the fake-outs, like old space junk and satellites that don’t work anymore.

Also Read: Variable stars might be the answer to our search for extraterrestrials

Researchers trying to rope in big space agencies

So, Vasile and the gang are suggesting setting up a system to process UFO pictures using fancy computer tricks. To start, they need a big collection of data about stuff in space, like satellites and all kinds of objects. This even includes space junk! So, they’re talking about grabbing data from NASA, the European Space Agency, and other groups around the world. This data needs to cover a lot of different situations, like where things are in orbit, how they move, and what they’re made of.

So basically, scientists need a solid database with info on all the stuff we’ve put in space to tell if something is really a UFO or not. But since a lot of this data isn’t easy to get, Vasile and his crew came up with a smart computer program that makes fake training data for the machine learning system. Then, they did a two-part thing to figure out what materials are in a spectrum. One part used fancy machine learning, and the other used good ol’ math to find the best fit for the data.

How did Machine Learning come in handy

After that, they used a machine learning system to guess how likely it was to find a mix of materials in a certain group. With all that computer stuff done, Vasile said they put their system to the test, and the results were pretty promising.

“We ran three tests: one in a laboratory with a mockup of a satellite made of known materials. These tests were very positive. Then we created a high-fidelity simulator to simulate real observation of objects in orbit. Test were positive and we learnt a lot. Finally we used a telescope and we observed a number of satellites and the space station. In this case, some tests were good some less good because our material database is currently rather small.”

In their next paper, Vasile and his team are going to talk about how they figure out the way objects are facing using their system. They plan to share this at the AIAA Science and Technology Forum and Exposition happening from January 8th to 12th in Orlando, Florida.