With the news that South Korea discovered some suspicious camera-equipped spy drones flying near its border, we’d imagine that the country might want to develop ways of countering that technology. Firing guns or missiles would be an option, of course, but another way of taking on enemy drones is with drones of your own.
Funded by a South Korean defense research institute, a group of roboticists at the Korea Advanced Institute of Science and Technology (KAIST) has been testing out ways of using autonomous unmanned aerial vehicles (UAVs) to locate, intercept, and disable other UAVs.
One of the drones used by the Unmanned System Research Group at KAIST. In this photo, the drone is carrying strawberries as part of an event in which students could order fruit to be delivered to them on campus. In another demonstration, the drone was used to transport a ground robot designed to attack enemy positions.Image: KAIST USRG
Dr. David Shim, who leads KAIST’s Unmanned System Research Group (USRG) and also the Center of Field Robotics for Innovation, Exploration aNd Defense (C-FRIEND), explains what prompted this sort of research and testing:
We imagined in near future there would be UAVs fighting UAVs. […] As found in many cases including the recent incident of [a DJI Phantom] wandering into the White House, even if you know UAVs are out there, it is very hard to stop them. One can try to shoot them with rifles or missiles, but they are too small for guns or guided weapons. So our solution is to stop them with another UAV, as they say eye for eye, and tooth for tooth.
Shim and his group are working with a variety of UAVs with different capabilities. These include agile multi-rotor UAVs equipped with nets that can be dropped on enemy UAVs to disable them. He says the biggest challenge is in programming the drones to operate fully autonomously, using onboard vision to detect the target UAVs, and then bringing them down by precisely releasing the nets on them.
The goal of the project, which is still in its early stages, is developing UAVs that could be used as part of an anti-drone defense system, and that could also go on the offensive if necessary. For a recent demonstration, Shim envisioned a scenario in which his UAVs must take on a rocket-launching enemy vehicle, which is itself guarded by its own UAVs.
KAIST researchers demonstrated attack UAVs confronting an enemy vehicle.Image: KAIST USRG
For the test, the first UAV to take off was a fixed-wing “eye-in-the-sky” reconnaissance drone, used to gather intel on the enemy. Next, a swarm of small, agile UAVs took to the sky. These small UAVs had two tasks to perform: neutralizing the guard UAVs and also escorting a larger attack UAV (which in a real conflict would transport a small ground robot with an explosive charge).
Here’s video of the test (note the “enemy vehicle” cutout on the lower right, and keep in mind that the UAVs are fully autonomous from takeoff to landing):
Originally, the idea was to get all the attack drones to fly in formation, cooperatively carrying one giant net that they’d use to sweep up the guard drones like a fishing boat does with a drag net. However, GPS wasn’t sufficient to keep the drones in formation, and high winds meant that the drones instead got one smaller net each, attached with magnets so that the net separates from the attack drone when an enemy drone gets tangled in it.
Dr. Shim says he expects to see more drones using a net (or a wire) to intercept other drones. In fact, a Dutch company called Delft Dynamics recently demonstrated a system that can fire a net from a cannon on one drone to snare another:
It’ll be interesting to see how the guard drones start evolving to combat techniques to bring them down, whether it’s attempting to avoid capture, flying stealthily to avoid detection, fighting off the attacking UAVs by dropping nets on them, or perhaps figuring out ways to defeat nets entirely.
In the meantime, Dr. Shim is happy to offer “the White House folks a few UAVs like ours when they spot another UAV over their perimeter again.”
[ USRG ]
Updated 2 April 2015 12:14 p.m.