navigation-background arrow-down-circle Reply Icon Show More Heart Delete Icon wiki-circle wiki-square wiki arrow-up-circle add-circle add-square add arrow-down arrow-left arrow-right arrow-up calendar-circle chat-bubble-2 chat-bubble check-circle check close contact-us credit-card drag menu email embed facebook-circle snapchat-circle facebook-square facebook faq-circle faq film gear google-circle google-square googleplus history home instagram-circle instagram-square instagram linkedin-circle linkedin-square linkedin load monitor Video Player Play Icon person pinterest-circle pinterest-square pinterest play readlist remove-circle remove-square remove search share share2 sign-out star trailer trash twitter-circle twitter-square twitter youtube-circle youtube-square youtube

DoD weapons designer: Swarming teams of drones will dominate future wars

March 30, 2017 (Photo Credit: DARPA)
WASHINGTON – Future wars will be fought with swarms of expendable, disaggregated, intelligent systems rather than the big, expensive weapon platforms the U.S. has relied on for fifty years, a top Pentagon weapons scientist asserted this week.

William Roper, the lead of the Pentagon’s semi-secret Strategic Capabilities Office (SCO), made his prediction March 28, adding that he believes the Air Force will have a greater challenge adjusting to this new reality than the other services.

“What used to be solo systems are going to have to be teams to be relevant in the near future, maybe even the far future. The technology is available today to make teams of systems higher performing than solo systems can be on their own,” Roper said at an event hosted by the Air Force Association’s Mitchell Institute. 

The idea of swarming systems being paired with a major system in a “loyal wingman” configuration isn’t a new one. The Pentagon’s various research groups, including DARPA and the service research labs, have been playing with the technology for years. 

But Roper’s vision of the future emphasized not just the capability, but a strategy behind it – that these systems will be cheap and disposable, something that in his words “you can simply throw away” without concern. It’s a situation he compared to the cultural move away from bringing nice china on a picnic in favor of paper goods that can be crumpled up and tossed after a meal.

“What this layer is giving us, this unmanned team with a manned system, is really just giving us expendability. We have not had that in our platforms, maybe ever,” Roper said. 

“All the things we build are expensive and if they take off, it’s our expectation they come home and land. That hasn’t been an issue until now, but if you think about highly competitive peer-on-peer war, how you have that system come back and land is a huge constraint,” he continued. “You have to protect it, you have to refuel it, you have to get them and maintain it and sustain it.”

The end goal, in Roper’s mind, is to limit the danger to individual operators. Rather than send in a wave of manned planes for the first day of combat, the SCO head said, send in wave after wave of cheap, disposable systems that come with no risk of losing a U.S. servicemember.

Michael Blades, an analyst with Frost and Sullivan, says the pieces for making the loyal wingman concept work are all available – but getting them to talk together is the tricky part.

“The technology is there but it all has to work together, and it has to work almost flawlessly,” Blades said. “We can't even get a fighter jet to work for all the requirements of the different services, and we can't decide what capabilities the MQ-25 should have, so do we really think we are ready for AI-controlled loyal wingmen?”

The big hurdles, Blades said, will be with datalinks and cybersecurity, something Roper acknowledged could be a challenge that needs to be worked out.

Michael Horowitz, a former Pentagon official now with the University of Pennsylvania who has studied drone issues, says that another challenge is the uncertainty around how technology will develop.

“Even within the AI research world, there is disagreement about how quickly various advances will occur,” Horowitz said. “Given that AI has utility far beyond the military context, commercial and academic research alone will likely push the realm of the possible in autonomous systems forward.” 

Cultural Pushback?

To Roper, the future of drone swarm warfare will occur across all the services. But he is concerned that the Air Force may not be as willing to take part as other services, in part because they have no faced the same kind of fundamental challenges the other services have. 

“The worst thing is to be in a position where the need to change is, yes, the most important thing, but it never the most important thing today,” Roper said. “We have to keep urgency in the need to change, because when we’re goanna need those changes, [when] the country says I now have a need for a next generation Air Force, it’s too late.”

Ironically, two of the most prominent AI/swarming programs from SCO are tied to the Air Force.  One, known as Avatar, is focused on the loyal wingman theory, taking a manned jet and having several unmanned systems tied together under its command. (There is also a ground-based equivalent being worked for the Army, Rpoer said.)

And perhaps the most attention-getting program from the office has been the Perdix system, a series of small drones designed with off-the-shelf technology. Perdix can be launched from the flare dispensers on an F-16 or F/A-18, and then can automatically link up and form a self-driving swarm of small ISR devices. A test in October successfully showed that the system could link 103 of the drones together. 

Roper expressed surprise at the attention Perdix has gotten, saying he expected more interest in the work the office had done on other programs.  But the lessons learned from testing the system, including the software that powers the autonomous swarming capabilities, could be applied towards future swarming programs, he noted.

Horowitz notes that it is a “challenge” for any service to incorporate “emerging technologies that could threaten established ways of doing business,” which leads to a prediction that AI will make its way into existing operational concepts before we see anything revolutionary. 

Adds Blades, “It's a lot to execute in the implied short time, but with the rapid advancement of many technologies I think we could see things like AI and loyal wingmen integrated into normal ops, piecemeal, starting in 10-15 years.” 

Next Article