Enclosed spaces, from tunnels to caverns to the interior of buildings, are poor fits for aircraft. Even relatively stable vehicles, like four- or six-rotor drones, still risk weird aerodynamics and can abruptly crash into people, walls, or anything inside. But there’s a compelling reason to at least consider drones that can fly inside buildings: that’s where the people are.
According to The Times, Strategic Command in the United Kingdom is working with an unnamed British company to develop a 3-foot long, six-rotor drone, designed for urban conflict and named, simply, the i9.
“It is the UK military’s first weaponised drone to be able to fly inside,” writes The Times, “using a combination of physics and AI that allow it to overcome “wall suck,” which causes drones with heavy payloads to crash because of the way they displace air in small rooms.”
Developing robots that can safely and effectively navigate inside buildings, and especially in tunnels or caves, is such a tricky problem that in the United States, DARPA set up an entire Subterranean Challenge circuit. For the Subterranean Challenge, multiple teams built robots and programmed AI to navigate and locate special markers, like a cell phone, gas leak, or a mannequin with a speaker, and were graded on performance.
A recurring theme among teams that competed in the Subterranean Challenge was the difficulty in controlling the robots. Cave walls are especially difficult for transmitting radio signals, but the rebar-and-cement of urban environments can pose a similar problem. In the DARPA competition, many teams addressed this by having the robots drop signal relays, and by leaning heavily on autonomous navigation.
Autonomy can solve the navigation problem, in part, but it also means that humans operators lack direct control over what the robots are doing. In a rescue mission, that is less of an issue, as robots are mostly equipped only with cameras, lights, and transmitters.
It is a much different ask if the drone in question is armed.
“Fixed with twin stabilised shotguns, [the i9] is also expected to undergo trials with other weapons including rockets and chain guns,” reports The Times. The i9 hexacopter is absolutely built to carry weapons, and to use them.
Initial targeting by this drone will come from machine vision, or onboard image processing software that interprets what the camera sees, and then in this case pairs that processing with AI for target tracking and stabilization.
Present rules of war, as well as best practices by many modern militaries, require that if a robot is going to fire at another human being, there is still a “human in the loop” to make that decision. The i9 is no different, and The Times says that a soldier would use “a live feed on a screen to decide whether to open fire.”
That keeps a human in the loop, but it only does so if the drone can effectively communicate with the human, which is especially challenging when operating indoors. This is true even if the soldier and the drone are in the same building, and it is especially true if the soldier is outside the building and directing the drone, which is the whole ostensible purpose of mounting twin shotguns on a drone.
All of this, also, is to say nothing of the effect of recoil on the drone’s ability to keep flying. While there have been drone-mounted guns and rockets before, most are designed to fly in the open air, where there is space for recoil to move the drone backwards safely. If firing shotguns or, as proposed, rockets or even a chain gun indoors, it is highly likely that the drone crashes into a wall or some other interior fixture.
In the best-case scenario, the human responsible for the drone can turn off the weapon when the drone crashes. Given the difficulty of signals traveling inside buildings, that may not even be possible.
The stated mission of the i9 is to give the British Army another tool for fighting enemies inside barricaded buildings. Even compared to other ways of putting weapons on a drone, shotguns and indoor flight are overkill, with dubious benefits, especially when compared to other options, like drones that can drop a grenade on the barricade. There is also the whole category of small loitering munitions. These are backpackable missiles like the Switchblade that soldiers can pilot into a building, confident that because it is a one-way explosion, using it is like firing a missile, instead of trusting an AI with a chaingun.
It remains to be seen if the i9 will deliver the capabilities it promises, without the foreseen complications. There is a chance that the designers have solved the problem of communications inside buildings, in a way that DARPA competitors have not, and one that satisfies requirements for meaningful human control over semi-autonomous weapons.
Until then, I fear “wall suck” is not the only internal hazard facing the i9 drone.