Evan Gretok, a PhD student at University of Pittsburgh, presented his work on a space computer at the Small Sat conference this week.
STP-H6-SSIVP was designed and built by the Center for Space, High Performance, and Resilient Computing, or SHREC. It was attached to the International Space Station from 2019-2022 and was designed to test supercomputing in space. Gretok says his work was to use artificial intelligence to train the satellite to autonomously identify and classify satellite imagery.
“So if you've ever played in Google Earth, or looked at images from satellites, usually it's a huge image and you have a ton you could potentially be looking at. What I do is split it into these individual tiles, and I teach a rudimentary computer brain how to classify all these different tiles, how to split them into groups,” said Gretok.
Gretok said he sees a lot of potential uses for this type autonomous learning.
“We’re currently dealing with climate change all over the world. We’re having in some situations, crop shortages and food scarcity," Grektok said. "There are actually organizations out there that take these satellites and the data coming from them, and actually can figure out what the soil moisture is, what the temperatures are, and find exactly what characteristics are right for certain crops, and actually put boots on the ground to go and talk to farmers. And say, like, ‘this is how you can actually use satellite data to feed your family.’ That's the kind of thing that I really see being one of the most impactful areas that this kind of system could be useful for."
SHREC is led by the University of Pittsburgh, but Brigham Young University in Provo, and the University of Florida and Virginia Tech are all contributing members. They work together to solve space flight problems and study space computing for space science, earth science, and defense.
