posted on Dec, 16 2020 @ 11:14 AM
In a recent test by the Air Force, a U-2 with the ARTUµ system installed successfully found simulated missile launchers, while the AI system ran the
aircraft sensors. The system took control of the on board sensors after takeoff, and hunted for the launchers, while the pilot monitored the sensors
watching for hostile aircraft. The system had been put through half a million simulated training iterations prior to the flight. The AI also
performed tactical navigation over the simulated battlefield, and was able to switch the radar system between search and self defense, without relying
on the pilot to override it. The system is based on the µZero program that is used in Go, chess, and video games.
ARTUµ appears to be related to the Agile Condor that was first delivered in 2016. Agile Condor used onboard computing, and machine learning to
assist with detecting targets of interest. It was flown on an MQ-9 Reaper, and was expected to be integrated on the Reaper in 2019.
www.af.mil...
taskandpurpose.com...
www.popularmechanics.com...
www.uasvision.com...