Space agency leaders discuss ever-expanding roles for AI


BREMEN, Germany – A European Earth observation cubesat scheduled to launch in 2020 will demonstrate an important space application for artificial intelligence.

One of the two six-unit cubesats that comprises the FSSCat mission will rely on artificial intelligence to determine which images to send from the satellite to ground stations. “For instance if you have clouds, it makes no sense to send normal pictures,” Jan Woerner, European Space Agency director general, said Nov. 19 at the Space Tech Expo here.

Polytechnic University of Catalonia and Estonian start-up Golbriak Space OÜ won ESA support for the FSSCat mission during the 2017 Copernicus Masters Competition. FSSCat, which stands for Federated Satellite System 6U tandem mission for sea ice and soil moisture monitoring, also won the 2017 Sentinel Small Satellite Challenge for its plan to gather data for the Copernicus Land and Marine Environment services program with two microwave instruments and a multispectral sensor on each satellite.

At the Space Tech Expo, space agency leaders said artificial intelligence will play key roles other programs, as well.

To prevent collisions in orbit, for example, spacecraft will need to rely on artificial intelligence, Woerner said. When collision avoidance maneuvers are conducted manually, the satellite moved often is no longer in its optimal orbit.

“After the collision avoidance, the satellite should be on the best path to give the data in the future,” Woerner said. “If we are looking to have megaconstellations in the future, then we for sure need artificial intelligence.”

In addition, space agencies and companies are likely to turn to artificial intelligence for proximity operations with cooperative or noncooperative objects, said Giorgio Saccoccia, Italian space agency president. Satellite operators will employ artificial intelligence when they need to move close to a satellite to perform maintenance or to move it out of orbit, he added.

Artificial intelligence also will continue to play an important role in helping companies and government agencies make sense of all the data gathered from space.

In Earth observation, “we are getting down pentabytes of data,” said Pascale Ehrenfreund, chair of the executive board of the German space agency DLR. “You wonder will they ever be really used?”

Startups may help answer that question by applying machine learning to the data to solve various problems, she added.

In many cases, the startups will combine the Earth observation data with information from other sources, including text messages, Facebook or Twitter, to understand “what is happening at the specific point,” said Yoshiaki Kinoshita, director of the Paris office for Japan Aerospace Exploration Agency.

Some of the startups creating applications that draw on data from Europe’s Galileo and Copernicus spacecraft already employ artificial intelligence, said Jean-Yves Le Gall, president of the French space agency CNES. “I am sure what we see today for Galileo and Copernicus in a few years we will see for the lunar economy. When we have infrastructure developed by the government, startups will propose applications using the infrastructure and applying new technologies like additive manufacturing and artificial intelligence.”