Intel Heads into Space with World’s First AI-Loaded Satellite
So, on 3 September 2020, Intel officially became one of the first AI technology enablers to supply its VPU in a “satellite-as-a-service” space mission.
From locating underground minerals to tracking flight course in the air and practically everything else in between them, Artificial Intelligence (AI) has become a ubiquitous element of our lives, transforming every bit of our lifestyle with guaranteed convenience and ease of work. But, now, Intel is taking things to space with its latest ‘AI in a satellite’ project. Intel’s most advanced AI-based chip has made it to space with the launch of PhiSat-1, an experimental satellite launched into the orbit on 3 September 2020. PhiSat-1 carries the Intel® Movidius™ Myriad™ 2 Vision Processing Unit (VPU) in its miniature configuration which is almost the size of a regular cereal box. The satellite was ejected from a rocket’s dispenser along with 45 other similarly small satellites. As of now, it is soaring at over 17,000 mph (27,500 kmh) in sun-synchronous orbit about 329 miles (530 km) overhead.
Intel® Movidius™ Myriad™ 2 Vision Processing Unit (VPU) — the same chip that made it to the space is the same component that you might find inside many smart cameras and drones. Imagine, jsut how a $99 USD selfie drone with an AI-loaded VPU could also become an inspiration for space technology providers in the future?
What is PhiSat-1?
PhiSat-1 (Φ-Sat-1) is the most-advanced nano-satellite to carry onboard an artificial intelligence VPU to improve the quality of geospatial terrestrial data. Intel’s Movidius Myriad is just one of other advanced nanosatellite components that made it to space. PhiSat-1 is one of the two CubeSats that were flown into space to collect Earth data related to polar ice and soil moisture as part of the FSSCat (Federated Satellite System) mission – a Copernicus Masters winning idea.
You can learn more about PhiSat-1 satellite here.
How is Intel’s AI VPU Improving Satellite Performance?
Intel is helping PhiSat data team deal with the large volume of Big Data generated by the high-fidelity cameras attached on board. Gianluca Furano, data systems and onboard computing lead at the European Space Agency, explained the real challenge in working with data generated from the hi-fi cameras.
“The capability that sensors have to produce data increases by a factor of 100 every generation, while our capabilities to download data are increasing, but only by a factor of three, four, five per generation,” says Gianluca Furano.
For the past 3 years, Intel, the European Space Agency and Ubotica have been working on an AI-powered cube satellite that can solve a challenge the space industry has had for many years: many of the images collected from outers pace and sent back to Earth are of clouds, so not useful at all. With roughly 70% of the images being of clouds, it’s a big waste of bandwidth and researcher time. But, thanks to the Intel Movidius VPU inside the recently launched cube satellite, now only useful images are being sent down to Earth. The really exciting part is that now that the AI satellite is proven to work, Intel and ESA are hard at work on a 2nd gen that will be trained to do things like detect forest fires, track deforestation, find planes that go down in the ocean, etc.
Gianluco added, “And artificial intelligence at the edge came to rescue us, the cavalry in the Western movie.”
The idea the team rallied around was to use onboard processing to identify and discard cloudy images — thus saving about 30% of bandwidth.
Intel closely worked with Ubotica, an Irish space technology startup that built and tested PhiSat-1’s AI technology.
Aubrey Dunne, chief technology officer of Ubotica said, “The Myriad was absolutely designed from the ground up to have an impressive compute capability but in a very low power envelope, and that really suits space applications.”
Here’s an explainer video from Intel: