Intel and ESA Launches First AI-Enabled Satellite, FACET-1


  • This week the European Space Agency (ESA) and Intel announced that they had successfully placed the first satellite into space, Fiat-1, with on-board AI-processing.
  • The PhiSat-1 uses Intel’s Movidius Myriad 2 chip, which was not originally designed for space travel.
  • PhiSat-1’s on-board AI is capable of selecting and automatically removing images of Earth if they are too obscured by a cloud.
  • AI has become mainstream on Earth, but getting it on satellites is a very big challenge, and PhiSat-1 can pave the way for innovations in how satellites detect natural disasters, or even us Mars How to communicate with Rovers.
  • For more stories visit the Business Insider homepage.

On 2 September, a satellite the size of a cereal box sailed for space.

Named PhiSat-1, its mission was to monitor the polar ice and soil moisture from which it was made – at least superficially – a fairly obscure piece of kit.

But for the satellite’s creators – the European Space Agency (ESA), chip giant Intel, and Irish robotics company Eubotica – this launch represents months of work, and has been postponed by a failed rocket launch, two natural disasters and a global epidemic. was given.

It was also a big technological leap.

PhiSat-1 is the first satellite to make it into orbit with AI onboard – and technology can change the way it responds to disasters, such as oil spills and wildfires.

Why has it taken so long to bring AI into space?

Gianluca Furano of ESA told Business Insider that although AI was popular for manufacturing products on Earth, getting technology on satellites has been a major challenge.

“Using AI in data-critical applications is also not straightforward on Earth,” he said. Flying an AI-powered chip such as the PhiSat-1, a ship 329 miles up on the planet, means any repairs, software patches, or upgrades will be a lot more difficult.

“Complexity just bursts,” Furano said.

On top of that, radiation into outer space makes it harder to build an AI chip that can withstand being in orbit.

“Whatever is silicon inside is disturbed by ionization radiation,” Furano said. This means that computer chips that work perfectly on Earth have a significantly higher error rate, and can also catch fire if they are sent to an external location without shielding or modifications.

But too many satellites carry custom-made spaceworth chips, the PhiSat-1 uses Intel’s myriad 2 chip, a handful of commercial drones in a commercially available chip – and even Magic Leap’s AR Goggles There is also a headset for.

Intel AI Satellite 3

The Fiat-1 processing unit depicted here uses Intel’s numerous 2 chips.

Tim Harman / Intel Corporation


Ubotica took Myriad 2, adjusted its software, and was able to turn off some electronics if it was likely to overheat. The researchers tested it by taking it to Earth’s largest particle accelerator in Cern, Switzerland.

By May 2019, the satellite was ready, but took longer than planned to make it into space. The original launch was scheduled for September 2019, but the rocket meant launching the Fiat-1 failed to launch, prompting an investigation.

While he was walking, the coronovirus epidemic set in – and to shut it down, two hurricanes came to the launch complex in French Guiana and a ground station in South Korea.

On 2 September, Fiat-1 finally made it to space, and has successfully circled it since putting its AI into practice.

Fiat-1’s AI lets it ignore the clouds – which is a lot more impressive than it sounds

Like the satellite, Fiat-1’s on-board AI function doesn’t seem glamorous at first, as it essentially auto-deletes photos of the Earth, making it much worse.

Before sending PhiSat-1 pictures back, its AI decides if they are sufficient for cloud cover. If the AI ​​obscures more than 70% of the photo by the cloud, it automatically scars the image.

This is useful because it saves a lot of energy to FIJAT-1. Aubrey Dunne, chief technology officer at Eubotica, told Business Insider that about 67% of the Earth’s surface is normally covered by clouds, so that means AI is cutting out a lot of images, rather than using processing power To send idle images back to scientists.

“Cloud detection is about data reduction,” he said.

It can pave the way for fighting fires and driving Mars rovers

Furano said that while cloud-scraping AI is exciting with the successful launching of FASAT-1, the technology’s current applications are “low-hanging fruit”.

“The real exciting part comes when you think about implementing onboard processing with machine learning [the type of AI PhiSat-1 uses] For applications where you need a very quick answer, ”he said.

As an example, Furano said that AI can rapidly increase satellite alerts of disasters such as oil spills.

Aubrey Dunne of Eubotica said the company was already working on the technology to build satellites that could automatically detect wildfires and flares from oil refineries. “We’re trying to reduce latency, so we’re using AI to do applications where timing is really important to get answers, and fire detection is a great example,” he said.

California wildfires

Ubotica is developing technology to help detect and track satellites automatically.

Via Josh Adelson / AFP Getty Image


“You want to try to alert authorities and concerned people at relevant locations about the location and extent of the fire, and how the data is changing without having to wait a day to download and wait,” And transfer is taking place. It’s day to act on the ground, “he said.

Dune feels that AI may also have off-world applications, for example to speed up communication with the rover robot on Mars.

UBOTICA aims to get the next innumerable AI chip in early 2022. “It’s a few satellites about twice the size of a Fiat satellite, and it’s partially ESA-funded, but partially commercially funded, so it’s a good type of bridge. Also on the commercial side. , “Dunne said, although he did not elaborate on where the commercial money would come from.

Jonathan Byron, head of Intel’s Movidius technology office, told Business Insider that Intel sees AI satellite chips as a nascent market, but one that promises to evolve.

“The technology is 20 years old which is on satellites, and 20 years ago GPUs were starting to grow massively – and you’re starting to see the same thing,” Byrne said.

Furano said that the technology currently in Orbit is 15 to 20 years behind the processing power that can be found in smartphones today, meaning new, improved chips top their list. “We are in dire need of AI in space,” he said.

Leave a Reply