The Office of Naval Research (ONR) recently concluded a competition to develop open-source software algorithms that could work in various drones.
Called the Artificial Intelligence for Small Unit Maneuvers Prize Challenge, the contest was open to the defense industry, academia, small and large businesses, and start-ups.
Artificial intelligence must enable unmanned systems to operate smoothly in a GPS-denied environment and perform autonomous navigation, object recognition, and mapping.
After three phases conducted over 10 months, California-based EpiSys Science won, taking home $750,000.
The winning company could enter into an agreement with the Department of the Navy to further develop the software and field an operational prototype.
“ONR can see if the software algorithms can work as promised across different drones — and contestants experience some of the obstacles that warfighters face in the field,” program officer Dr. Michael Qin said.
“The technology leaves the lab and faces the noise and ambiguity of the real world.”
Competition Phases
All participants were required to submit white papers and present their ideas virtually as part of the competition’s initial phase.
Qualified teams were then tasked to compete in a virtual environment and create specific software algorithms for simulated scenarios.
The third stage of the competition allowed participants to install their algorithms in actual government-furnished drones and participate in a live exercise.
The drones with artificial intelligence must identify objects, maneuver indoors, and create a map of their specific location.
“To ensure our nation’s warfighters maintain superiority, we must harness AI-powered autonomous systems, from large-scale drones to the smallest micro systems,” senior program analyst Kelly Hughes explained.