As the Pentagon moves forward with tests on autonomous drone swarming technology, few doubt that the warfighting concept will work.
But major questions persist, such as how to classify drone swarms and just how smart to make the network that allows hundreds of little autonomous machines to communicate, the Assistant Secretary of the Air Force for Acquisition, Technology and Logistics told reporters at the Pentagon on Wednesday.
“How do you certify [drone swarming]? How do you test and evaluate it? And who owns it?” he said. “Is it a weapons system? Or does the platform using it own the autonomy and swarming and collaboration? Or is there a program … that plugs it into all sorts of platforms?” Dr. Will Roper said.
Multiple efforts to develop drone-swarming technology are now underway inside the Defense Department.
The Defense Advanced Research Projects Agency, or DARPA, and the Air Force Research Lab have been working on a project called Gremlins: controlled micro-drones that will drop out of cargo planes to swarm enemy defenses, going ahead of fighters, ships or ground vehicles. AFRL and DARPA anticipate a flight demonstration sometime this year.
In January 2017, the Navy tested remote-controlled micro-drones dropping out of fighter jets. The exercise, held at China Lake, California, used three F/A-18multi-role fighters that launched 103 Perdix micro-drones from “specially designed canisters affixed under the aircraft’s fuselages,” The Washington Post reported at the time.
The Air Force has simulated a similar drop. In 2014, Perdix drones were dropped from F-16 Fighting Falcons at Edwards Air Force Base, California, the Post said.
Most recently, DARPA tested how drones could work together in a GPS-denied environment. In November, the research arm tested multiple drones at Arizona’s Yuma Proving Ground. The test showed the drones “efficiently shared information, cooperatively planned and allocated mission objectives, made coordinated tactical decisions, and collaboratively reacted to a dynamic, high-threat environment with minimal communication,” according to a news release.
But should the Pentagon worry about making the drone swarms too smart?
Roper said that while he believes that swarming technology will have a significant role in the future battlespace, the artificial intelligence must be designed carefully to be able to “contain the types of effects … so that it doesn’t do things that are unintended while still allowing it the freedom … without having to micromanage it.”
That applies to A.I. that learns and improves after each military operation, he said.
Brainstorming these ideas has forced officials to think beyond the traditional ways the Pentagon looks at potential programs, Roper said.
For example, while testing Perdix in his previous post as head of the Defense Department’s Strategic Capabilities Office, it “was tough to find a range” to define the boundaries for the smart technology, he said.
“I went to [leadership] and said, ‘Hey, I’d like to throw 100 micro-UAVs out of fighters.’ And they said, ‘OK, great, tell me the flight plan for each one.’ Well, I don’t have one,” Roper said.
He continued, “They’re going to do their own thing. But I can draw a box and make sure they don’t leave that box. And that’s just an example of how we have to … require different kinds of thinking.”
These types of technologies “are about infrastructure, data management, and it’s something that is more of like a network that goes across systems than something you look at,” he said.