MOUNTAIN VIEW, California –SES is working with the Massachusetts Institute of Technology to explore ways to use artificial intelligence and machine learning to simplify operation of its communications satellite fleet.
“We have a very large fleet and tens of thousands of telemetry signals on each of our satellites,” Valvanera Moreno, SES system architecture and innovation manager, said Oct. 10 at Satellite Innovation 2018 here.“The next satellites will have even more data to process. That’s why we think this area has a lot of value.”
Like SES, government agencies and space companies are applying artificial intelligence to various problems they face.
Orbital Insight, a geospatial analytics company, relies on artificial intelligence to help answer questions its customers ask.
“Artificial intelligence enables human analysts to extract maximum value from imagery,” said Devin Brande, Orbital Insight advance programs director. “We are on the cusp of combining modern remote sensing with other sources of intelligence to create a rich picture.”
Raytheon’s Intelligence Surveillance and Reconnaissance business established a capability center to focus its artificial intelligence and machine learning expertise. “As we grow that into a fundamental capability of our business, the goal is to dissolve the capability center and have it become part of the DNA of our business,” said Gabriel Comi, Raytheon Intelligence, Information and Services’ Artificial Intelligence and Autonomy Capability Center chief architect.
CosmiQ Works, the laboratory established by U.S. intelligence agencies to take advantage of the innovation of commercial space startups, holds competitions, called SpaceNet, that offer cash prizes to competitors who develop automated methods to detect road networks or other landmarks from high-resolution satellite imagery.
CosmiQ Works makes the winning algorithms open source. “Hopefully, that helps our government partners and the commercial sector,” said Adam Van Etten, CosmiQ Works technical director. “Sometimes these algorithms that get a lot of press don’t translate to our domain.”