Why the Flying IED Threat Has Barely Started
Last week, casualties from an explosive-rigged consumer drone in Erbil ignited new discussion about the flying IED threat. Writing for War on the Rocks, Ulrike Franke sees a manageable threat that is “unlikely to fundamentally change the fight.” T.X. Hammes paints a much darker picture in his response. He sees a disruptive technology that will “dramatically change the character of conflict between state and non-state actors.” So who is right?
We might start by asking a different question. Non-state actors have been flying consumer drones for at least a couple years. If Hammes is right, why have we seen so little weaponization? Where are all the attacks?
The short answer is that the flying IED revolution has barely begun because key technologies are not quite ready. Only two sectors of the civilian drone market currently offer high reliability: a handful of consumer video drones and high-end commercial systems. The optimum systems for flying IED attacks lie in the space between. As key technologies mature and reliable systems fill this space, expect to see flying IED attacks increase dramatically.
I know these technologies well because I spent two years founding and running Uplift Aeronautics, a nonprofit aimed at using drones to deliver humanitarian aid to besieged communities in Syria. We ultimately dissolved, but the experience of building and flying our own drones, writing custom autopilot software, and building partnerships in the consumer drone industry gave my team a world-class education in how the technology is evolving. Our paradigm was premised on using swarms of drones to push small amounts of cargo through contested airspace. We were acutely aware that our aircraft would make formidable weapons, so spent much of our time developing security features to prevent misuse. The learning curve we painstakingly climbed over those two years is the same learning curve that non-state actors are currently facing.
The easiest way to launch an attack is to strap an explosive to an off-the-shelf consumer video drone like the DJI Phantom 4. These are widely available and easy to fly but poor fits for weaponization. They are severely weight-limited, with a short battery life and limited range. A Phantom 4 weighs three pounds, so adding even a small payload pushes its limits. Similar constraints apply to almost any entry-level quadcopter or RC plane. The market trend is toward even smaller and lighter systems. Their polished proprietary interfaces and rich data feeds are wonderful for consumers but far from ideal for insurgents or terrorists.
At the other end of the spectrum are sophisticated commercial systems, but their high cost and low accessibility put them out of reach for most violent non-state actors. The exceptions will be large, well-organized, state-sponsored actors like Hizballah.
The real opportunity is found in the space between, for both good and ill. Custom-built drones in this space are larger, more capable, and more flexible than off-the-shelf consumer video drones, but significantly cheaper and more accessible than high-end commercial systems. My team’s cargo drones cost about $700 in parts, flew at 35 miles per hour, and could autonomously airdrop two pound payloads at 80-mile range (or four pounds at 40 mile range), usually hitting within 15 to 50 feet of the intended coordinates. Thanks to highly customizable autopilot software, it was trivial to turn off all data links, which rendered the drones immune to some electronic countermeasures. If the GPS was jammed, the drones could continue flight with at least some accuracy using a magnetic compass. We wrote software that generated semi-randomized flight plans to enhance swarm survivability. Our intent was to push humanitarian aid through contested airspace, but adversaries could use similar technology for attacks.
As a side project, I also experimented with building the cheapest “insurgent” drone I possibly could. The result required $4 of foam board, packing tape and hot glue, and about $250 in cheap Chinese components. It was ugly, but it could deliver two pounds at a range of six to 12 miles.
Emerging technologies will make these types of drones even more resilient. Optical flow navigation lets drones navigate by terrain, and rudimentary support exists for terrain following. The hottest new trend is computer vision, which, as Hammes notes, could allow drones to identify and track specific targets without external navigation aids.
The reason so few of these drones have been weaponized is simple: This is largely a DIY space, and reliable systems integration for drones is still very hard. Drone delivery companies have backed off from big promises and are quietly experimenting, learning, and iterating. One drone startup after another has failed or disappointed. My own team had countless crashes and a serious fire before we dissolved. Our experience was hardly unique. In a recent humanitarian drone challenge, ten world-class teams had several crashes, a fire, and various technical problems between them. This is the learning curve that drone entrepreneurs must climb, including terrorists and insurgents.
To get a reliable flying IED, you first need a vehicle. Many different kits can serve as the basic airframe, but these are rarely sold as complete solutions. Kits must be assembled and outfitted with various components like motors, speed controllers, props, and batteries, many of which come from China, are of uncertain quality, and can fail in myriad ways. Choosing the proper components and integrating them into a reliably performing aircraft is an enormous challenge. The problems compound at heavier weights. This means a less forgiving flight envelope, catastrophic crashes, higher current, more risk of fire, and more vibrations and stress. Getting a heavy fixed-wing drone airborne can be surprisingly difficult, and even a slight center-of-gravity error can send a drone cartwheeling on takeoff.
Various companies now sell complete solutions, but these are often indiscernible from hobby-grade products. Some companies simply prepackage foam hobby aircraft in standardized, validated configurations. Last year, Prioria Robotics was accused of selling hobby-grade aircraft to the Department of Defense for $240,000 apiece. Affordable, high quality systems are appearing, but the turbulent drone market is still sorting itself out. Reliability is still hit and miss.
Once you have a vehicle, it needs a brain. The most popular drone autopilot software stacks in the world are the open-source ArduPilot and its cousin PX4, both of which run on inexpensive open-source autopilots like the Pixhawk or various Chinese clones. This combination is powerful but difficult to master. These systems contain hundreds of parameters that can be individually tuned to affect everything from pitch and roll sensitivity to how the autopilot handles a GPS failure. Screw up any one of these parameters, and you may be in for a crash or a flyaway — a lesson we learned dozens of times. These are not problems you want when explosives are strapped to your drone.
Getting a swarm airborne is even harder. You need processes to safely store, transport, preflight, launch, and recover drones, which is more difficult than it sounds. For a successful flight to occur, hundreds of small details have to be perfect. Attending to all these details requires discipline, teamwork, and robust checklists and rules, just like in manned aviation. A team at the Naval Postgraduate School, which successfully flew a swarm of 50 fixed-wing drones, devoted much of its time over two years to these seemingly mundane details.
The rudimentary state of multi-vehicle communications technologies and ground control stations amplifies the challenges. The most readily available telemetry antennas and ground station software only work well with single vehicles. Getting swarms airborne is still possible — many research teams are doing it — but it is challenging, and requires expertise, practice, and a lot of hacking.
All of these challenges add up to a steep learning curve for non-state actors. Some groups have made the climb, most notably the drone-flying militias in Ukraine, which build and operate sophisticated aircraft using the technologies discussed above. Given their successful track record developing explosives and rockets, other groups will certainly follow.
However, if I am right, the situation is not quite as grim as Hammes predicts. At least, not yet. He rightfully notes rapid advances in 3D printing, and his points on lightweight EFPs are deeply concerning, but he downplays the difficulties of system integration. To paraphrase Clausewitz, everything with drones is very simple, but the simplest thing is hard. Some non-state actors may indeed develop the terrifying weapons Hammes envisions, but they will be the exceptions for now. I suspect most terrorists are still trying to evolve past flyaways and frequent crashes, like drone hobbyists everywhere.
The IED threat will grow rapidly as more reliable systems fill the void above consumer video drones. The requisite technologies are developing rapidly, and many new startups are working in this space. As better systems appear, simple deliveries of explosives will become much easier. The arrival of true multi-vehicle ground control stations will put swarm attacks within easier reach. Improved battery technology will equip drones to carry more weight over longer ranges.
We got a respite over the past year, because even as consumer video drones were booming, the DIY drone sector went through a convulsive collapse and rebirth. Market forces led Berkeley-based 3D Robotics, the company largely responsible for the rise of the DIY sector, to abruptly pivot away and stop developing critical autopilot hardware and software. This pulled the rug out from under much of the drone industry, forcing many companies to purchase stockpiled hardware from third-party vendors or use Chinese clones. Component reliability hasn’t exactly been improving. The industry is reorganizing and rebounding, however, and next-generation technologies are in active development. The reliability boom, and the rise of flying IEDs, still lies ahead.
What can be done? For starters, we should not overreact. We somehow manage to survive in a world where automobiles and human beings remain the most effective means of delivering large quantities of explosives. Drones and their enabling technologies are here to stay and have many positive applications. Overzealous efforts to throttle the technology now will only hurt American competitiveness in a fast-growing industry.
With that said, the threat of flying IEDs deserves serious and urgent attention. I can add little to Hammes’ excellent recommendations, except to say that there is no magic bullet. Data link jamming, GPS jamming or spoofing, hardening, and kinetic defenses all play a role, but all have potential weaknesses. The most effective defense will be a multilayered one. Defending against flying IEDs will be a cat-and-mouse game, just as it was with land-based IEDs, necessitating rapid innovation in technology and tactics. I am pleased to see large-scale exercises dedicated to these threats, as well as creative initiatives like a NPS/DIUx-hosted drone hackathon. We need such initiatives to retain the advantage.
We cannot afford to be complacent and should not infer too much from the relative absence of flying IED attacks thus far. Yes, small drones have proliferated rapidly, but the technology is still in its toddler years, and today’s widely available consumer drones are not ideal weapons. It is the next generation of drone technology that has me worried, and it will be here soon.
Maj. Mark Jacobsen (USAF) is a C-17 pilot and Political Science PhD candidate at Stanford University. He is the founder and former executive director of Uplift Aeronautics. You can read his other writings at buildingpeace.net. This article does not reflect the views of the U.S. Air Force or the U.S. government.