The small Perdix unmanned aerial vehicle, shown here, is rugged, allowing service members to toss it out the back of an aircraft and know it will work. Strategic Capabilities Office photo.
Designs for the seventh generation Perdix unmanned aircraft vehicle should be completed by this summer, William Roper, the director of the Defense Department’s Stra?tegic Capabilities Office, said in an exclusive interview with Air Force Magazine.
The body length of a sixth gen Perdix unit is 6.5 inches long with a wingspan of less than a foot, weighing in at just over 10 ounces with propellers that are 2.6 inches long. The expendable micro-drones are designed to be airdropped from various aircraft, so they can accomplish a variety of missions, including largely ISR objectives, and remain mutable for future payloads of varying degrees.
Roper said the program has progressed from one generation to another “roughly” every six or seven months, which he called “very good for the government.”
“It’s rare for us to build something in more of a smartphone-style fashion where you’re continuously improving the hardware and software,” he said.
SCO has identified a few areas for improvement in the next generation. For example, not all design facets are nailed down and the services have not put into stone what they want or need yet. Roper said he doesn’t ask partners to “sign their name on the line until they know what they’re signing up for.”
“I think we’ll have those designs done by the end of the summer,” he said. “And if seven is a lucky number, then that works out for everyone.”
The program could define the way forward for other SCO and DOD projects, from methodology to production to deployment, said Roper.
An Oct. 26, 2016 test showcased the Perdix’s ability to deploy from F/A-18 Super Hornets over China Lake, Calif., and organize into units of swarms, accomplishing various modes of organization. Though preli?minary results were released in January, “so much data” came out of the public test for the system’s sixth gen the team is still sifting through it eight months after the fact, R?oper said.
“It’s taken us a while to go through the data to make sure the auton?omy software works the way it’s supposed to,” he said. One of the ways the software is supposed to work during testing is in offering “self-healing” to swarms.
“We’re doing a lot of active things to the UAVs while they’re flying,” Roper said, explaining that the team wanted to see if after pulling a unit out of a group, “all of the others UAVs—all the other Perdix—recognize that and are able to communicate it to their brothers and sisters.” Another test was to determine whether the swarms could identify a returning unit that had been pulled out and reintegrate it into a mission set.
After months of data analysis, Roper said SCO is “very pleased” with the autonomy and swarming software.
“That was the real objective for the China Lake test—to make sure that we have, no kidding, made swarming software that allows it to work as a team without us telling it what to do,” he said. “So, check the box for that.”
How well did that test run? “Everything appeared to work according to plan,” Roper said, emphasizing this is, after all, the sixth iteration of the system. Initially developed in 2013, Perdix were first tested in 2015 by the Air Force Test Pilot School at Edwards AFB, Calif. “We weren’t as successful on the earlier versions,” he added.
Rugged good looks
The struggle in the beginning was taking commercial components and ensuring they could survive combat environments, including, moving at the speed of sound or getting dropped into temperatures as cold as -20 or -30 fahrenheit. While Roper said his team was able to test the swarming capability of the Perdix easily, by throwing them on poles and deploying them on the ground, it was the environmental testing he was more concerned about.
“It comes out of the airplane cold, it has to boot up as it falls. As it falls, it has to discover its orientation,” Roper said, adding the Perdix even has to figure out what it is, existentially.
Because Perdix uses commercial components, every one “is a little different,” he explained. So Perdix has to be able to determine its “abnormality,” for example, there may be a slight weight differentiation between each unit. The existing software allows it to adjust accordingly.
Though they are small, the “ruggedized” nature of Perdix, as Roper puts it, makes them appealing to services like the Marine Corps, with whom SCO ran the first test.
“It’s something you can just hand to a marine and have them throw out the back of a moving airplane and know it’s going to work,” Roper said. The program, however, is not just for the Marine Corps. The Air Force, Navy, and Special Operations Forces are also interested in the technology.
But what happens if the software fails? SCO is able to send kill signals to UAVs that aren’t responsive. During the China Lake testing, for example, if a Perdix wasn’t responding within 60 seconds, it got the kill signal, said Roper.
The program allows SCO a special methodology in stretching its reach since Perdix are expendable. Unlike many other DOD programs, Roper doesn’t have to worry too much about how to get the units back. This is one of the main pushes behind using commercial components for the unit exterior.
“There’s been so much commercial investment in miniaturizing, improving performance of processing, performance of battery, and making everything extremely reliable,” he said. “We’d be crazy if we put in government hardware because we have this big, mega-million dollar industry spiraling technology every other year.”
Even though a significant number of early Perdix failed to communicate back or were destroyed by the cold, during the China Lake test upwards of 90 percent of Perdix survived deployment.
“Now that we’ve [got] the discipline right, we can start thinking about payloads, which is the fun part,” Roper said. What will those payloads be? Well, “surveillance” is a no-brainer as Perdix is able to see things and get places Reapers and Predators cannot. And though ISR will continue to be a “staple” of the program, it won’t be the only angle. Roper said surprise is key to the military’s success in utilizing new technologies and so he was unwilling to share more specific mission objectives.
There’s only one Perdix in town
SCO aims to be able to upgrade Perdix by taking evolving tech as it emerges and “dropping it in.” This type of open architecture allows for mission versatility. When commercial technology improves every year, Roper argues, this method allows DOD to use it without rebuilding system from the ground up. But how long is Perdix aiming to last in a rapidly changing technological landscape
“When this transitions, the transition partner is going to have to manage it differently than traditional defense programs,” Roper said. “Rather than going out and buying 100,000 of these and having them sit on a shelf for 10 years, I think it’ll be interesting to see—can we learn to have small buys and continue to improve the product?”
The ability to change software rapidly, Roper argued, is the way of future warfare. Software that teaches itself is the future within industry. Companies like Apple and Google are championing this type of machine or deep learning. Not only is this the way SCO approaches many of its projects, but it may also be a capability built into the seventh generation of Perdix.
“Information is becoming so complicated,” Roper said. “We’re going to use Perdix as an example, as a pathfinder, to try to get that capability.”
While this will raise the challenge of the program, “I think it’s such a potentially transformative capability,” Roper said, it’s worth it. Roper couldn’t share what he wants to get Perdix to learn independently. But he did share that the final maneuver of the China Lake test, which was done autonomously by Perdix, was special. The software wasn’t built into the UAVs when the SCO team arrived at the lake. They identified the need while there and in a matter of hours programmed it into the units.
“I can’t think of another system that I’m aware of where you can change the software in the field,” he said. “We built Perdix that way.”