An Insitu ScanEagle is caught on the SkyHook retrieval system during a demonstration for leaders from the Coast Guard Acquisition Directorate and the Research and Development Center at a U.S. Navy testing facility in Dahlgren, Va., in May 2012. (U.S. Navy)
- Filed Under
On a wind-swept National Guard base near San Luis Obispo, Calif., at an outdoor testing facility, a group of researchers looks up at the sky. Circling above is a swarm of nine unmanned aerial vehicles, some red and some blue, like a multicolored flock of birds. On the ground, one of the researchers carries a 10th unmanned aircraft, shaped like a small stealth bomber, over to a set of PVC rails mounted at an angle. He turns on the UAV’s small motor and, using a long band of elastic, catapults it into the air.
A crowd of onlookers breaks into applause as another researcher steers the plane upward with a remote control, higher and higher, so the plane merges with the flying swarm. But members of the team appear too busy to notice. They continue to stare at laptop screens, monitoring the speed and altitude of each drone, while trying to ignore the computer voice chanting ominously: “Warning: loss of communications.”
This UAV demonstration, conducted in May at Camp Roberts, set a record for the Naval Postgraduate School’s Advanced Robotic Systems Engineering Laboratory, or ARSENL, part of the Consortium for Robotics and Unmanned Systems Education and Research. The effort was established in 2011 by Navy Undersecretary Robert Work to encourage widespread collaboration in the development and use of unmanned systems. In the previous field trial in February, the team flew seven drones at once. It’s all part of its ongoing campaign to gauge the potential utility of swarms.
Swarms are all the rage among UAV proponents, who tout their ability to overcome enemy air defenses, locate missing hikers and serve as surrogate cellphone towers, restoring communications after a national disaster. The idea has been around for more than a decade, but teams in government, academia and industry are just beginning to overcome some of the significant obstacles.
“UAV swarms can interpret what is happening in real time, self-organize and get the job done,” said David Scheidt, a member of the principal professional staff at Johns Hopkins University’s Applied Physics Laboratory and leading UAV swarm researcher.
STRENGTH IN NUMBERS
To be considered a swarm, UAVs must communicate with one another to perform tasks as an intelligent group, said Gabriel Santander, Boeing’s Advanced Autonomous Networks program director.
A swarm could, for example, conduct surveillance of an enemy compound. When a person emerges from a building, a single drone could follow that person and report his whereabouts. Other vehicles in the swarm could track additional individuals, produce detailed maps of the area, detect airborne chemicals and relay updates to a distant command post. As individual drones leave the swarm to fulfill their assigned tasks, the swarm could autonomously reorient itself.
Swarms also could fend off other swarms.
“We’ve witnessed a remarkable explosion of robotic unmanned systems,” says Timothy Chung, the Naval Postgraduate School assistant professor who established ARSENL in June 2012. Those systems are becoming increasingly capable and inexpensive, making them easy for potential adversaries to obtain in large numbers.
Even if enemy drones are not sophisticated, they might be able to overwhelm U.S. air defenses. It’s like a tennis match, Chung said. No individual high school player could possibly beat Swiss champion Roger Federer. “But if 50 high school players were lobbing balls onto his court, poor Roger wouldn’t be able to defend against that,” Chung said.
About four years ago, when Navy officials were discussing ways to counter the threat posed by swarms of unsophisticated UAVs, it occurred to Chung that defensive swarms might offer a less expensive solution than traditional air defense systems. “I don’t want to spend a million bucks to counter a $10,000 threat,” Chung said.
Cost is an important consideration for UAV swarm researchers, who want to buy drones in bulk.
“The real advantage of a swarm or any autonomous vehicle is that it accelerates the speed at which you can respond to things in the world,” Scheidt said. “With more vehicles, you can respond to more things.”
Swarm researchers also acknowledge that many UAVs crash. “There are two kinds of people doing this kind of research,” Scheidt said. “Those who say they have crashed UAVs and those who lie.”
Unlike the military’s multimillion-dollar drones used in Iraq, Afghanistan and Pakistan, future swarms are likely to be composed of inexpensive aircraft fitted with a single sensor designed to gather imagery or detect radiation, airborne chemicals or biological agents. Chung calls price “the elephant in the room,” adding that UAV swarms make sense only if each drone does not come with a hefty price tag.
Unlike state-of-the-art drones designed to evade radar and operate for days on end, the UAVs making up swarms are likely to be inexpensive and expendable. Resilience comes from the emphasis on quantity over quality.
“Think about a bunch of ants going after a potato chip,” Scheidt said. “You might step on a couple of ants, but the ants still get the chip.”
Commercial off-the-shelf drones equipped with GPS and autopilot often cost $10,000 to $15,000 each. While that may sound inexpensive compared with traditional military systems, the price is still too high for many drone researchers.
Naval Postgraduate School teams build their own hobby-grade drones, paying $200 for the polypropylene airframe with a wingspan of about 5 feet and $180 hobby-grade autopilot that features GPS and an inertial measurement unit. The total cost, including two batteries, onboard processers, antennas and airspeed sensor, still comes in well below $1,000 for each drone, Chung said.
The military can reap additional savings from the efficiency swarms provide, Santander said.
“When fielded, high-level autonomy technology will offer military customers valuable applications from intelligence, surveillance and reconnaissance to communications missions at a fraction of the current cost,” Santander said by email. “As processing and decision-making becomes autonomous and distributed, the ‘manned’ requirement in unmanned aerial system missions substantially decreases, lessening the sensor product currently overloading analysts.”
In field tests conducted in 2012, Boeing and the Johns Hopkins Applied Physics Laboratory demonstrated that a person equipped with only a laptop computer and tactical radio could guide a swarm of UAVs designed to autonomously search an area and obtain data from the swarm. That result is significant because it eliminates the need for a ground control network.
In 2011, Boeing demonstrated that a swarm featuring two types of drones — two ScanEagle unmanned aircraft, built by Boeing subsidiary Insitu, and a Unicorn, built by Lockheed Martin’s Procerus Technologies — could conduct a reconnaissance mission. During field tests in Oregon, the three drones searched an area, communicated with one another and reported their findings to teams on the ground.
THE BATTLE OF 100 UAVS
The Naval Postgraduate School is eager to conduct the world’s largest swarm demonstration, pitting rival swarms of 50 UAVs against each other. Chung and his colleague, Kevin Jones, Naval Postgraduate School research associate professor, plan to discuss plans for the 2015 competition in August at the Association for Unmanned Vehicle Systems International conference in Washington, D.C.
Still, the Naval Postgraduate School demonstration at Camp Roberts shows how difficult it is to prepare, launch, monitor and recover 10 drones, let alone 50. It’s never been done before, Chung said.
During the Camp Roberts demonstration in May, the Naval Postgraduate School’s six-person team was in a hurry to launch the 10th drone because each vehicle only has about 45 minutes of battery power. By the time the team launched nine UAVs and was preparing to send up the 10th, the batteries on the first aircraft in the fleet were nearly exhausted.
That exercise also demonstrated how hard it is to communicate with more than a handful of drones operating simultaneously. When the school’s team flies five at a time, the command-and-control link between the ground station and the aircraft goes haywire, sending a “warning: lost communications” message for each aircraft.
“Basically, all the aircraft are trying to talk to the ground control system at the same time, and if one can’t get any data through for a few seconds, it starts to complain,” Jones said by email. Eventually, school officials plan to employ a command-and-control network with more bandwidth.
They also would like to schedule incoming messages so the multiple aircraft do not all bombard their handlers with constant warnings, Chung said.
Swarm researchers also wrestle with software issues. Each UAV in a swarm needs to make decisions rapidly and prioritize tasks.
“It’s not artificial intelligence, but it does require advanced algorithms,” said Ray Buettner, the Naval Postgraduate School’s director for field experimentation.
Buettner oversees quarterly field-training exercises at Camp Roberts, known as Research and Experimentation for Local and International Emergency and First-Responders, or RELIEF, designed to help civil servants, military service members, academics and industry teams address common goals and challenges.
During the next field-training exercise, scheduled for Aug. 5-8 at Camp Roberts, Naval Postgraduate School researchers plan even more extensive UAV demonstrations. In addition to breaking their record by flying more than 10 drones simultaneously, teams are preparing to use drones to track moving targets and relay the information they gather through other drones.