War With Friends: Pentagon Eyes a Drone App Store

An MQ-1 Predator instructor with the California Air National Guard remotely controls his drone. He might control more than one, if a new Pentagon project comes to fruition. Photo: National Guard Bureau

The US military has dozens of different types of drones in its arsenal. Each one has its own unique controller. And each of those various controllers flies a single robot. There's no system that controls multiple drones at once. One Pentagon office thinks that's an archaic way of doing business.

Inside the Pentagon's Acquisition, Technology and Logistics directorate, a team is working on ways to operate different types of drones with a single controller. It's a big technical challenge ' one that's failed in the past ' since the different manufacturers of different drones each have proprietary control software. But the official in charge of the effort envisions a new drone software architecture that's agnostic about what kind of drone it controls; and allows human controllers to think in terms of drone fleets rather than individual robots, including fleets comprising different kinds of drones. That would enable a dramatic expansion of the possibilities of drone warfare.

Step one is to get a kind of universal remote for the drones ' that is, a controller that can operate, say, an armed Predator and a jumbo Global Hawk spy. It's a major challenge.

'The objective is to be able to 'shop' for mission specific applications and services from a single 'App Store',' says Rich Ernst, the Pentagon's lead officer for what's called the Unmanned Aerial Systems Control Segment, or UCS, in an emailed statement to Danger Room. 'The methodology is akin to the commercial 'smart-phone' industry, wherein applications are down-loaded to suit individual user taste and productivity. The repository allows small software businesses to compete on a level playing field' with the major defense conglomerates.

The first such company is California-based DreamHammer. DreamHammer has developed software that can operate numerous robots from the same tablet or laptop. Known as Ballista and first reported by Kashmir Hill at Forbes, it's a layer that sits on top of the proprietary software governing Predators, Global Hawks, and the rest of the military's robotic aviary, using application programming interfaces, or APIs.

DreamHammer's CEO, Nelson Paez, has brought Ballista to Ernst's attention. Ballista aims to be the first-ever multi-drone operating system ' basically, the universal remote. And that would cut through what Paez and Ernst see as an expensive irony. 'Currently,' Paez tells Danger Room, 'it takes more people to operate unmanned systems than it takes to operate manned systems.'

The user interface for Ballista, DreamHammer's multi-drone operating system. Image: DreamHammer

Consolidating the controllers has frustrated the military in the past. The Air Force has tried and failed to create control stations that are agnostic toward the types of drones they fly. Nor has it been able to control multiple drones from the same control station. All that has required a level of autonomy that the current drone fleet doesn't possess.

'It's like we're going to pull a rabbit out of the hat in the next three years in terms of automation,' a frustrated drone engineer told Defense News in 2009. The following year, the Navy quietly asked the defense industry if a 'common control system' for drones could even be built.

Ernst's job is to make that a reality. He wouldn't discuss how much money is going into constructing a common drone architecture, when it might come online, or even how many drones a single controller ought to control. But it appears that DreamHammer has an edge in this space.

Paez demonstrated Ballista to me a few weeks ago at a Washington, D.C. restaurant. He wasn't actually controlling any drones while drinking his coffee. But anyone using Ballista would be able to, using the display loaded onto Paez's MacBook. The majority of the screen is a 3-D rendered map showing a swath of territory that the drone flies over. Icons show the positioning of the drones, ostensibly in real time via GPS, with color-coded flight paths that a remote pilot can customize. Tabbed browsing shows layers with the mission plan, imagery recorded by the drone, additional aircraft nearby and other relevant data. Switching from one drone to another is a matter of a mouse click.

'Ballista needed the ability to fuse multiple dissimilar data sources (structured or unstructured, real-time or static) from any system into an open, consistent data model,' emails Chris Diebner, DreamHammer's chief technology officer and one of Ballista's main designers. Efforts over the course of months by Danger Room to see Ballista in action using Navy drones were unsuccessful.

Ernst declines to talk about Ballista, or any competitors it may have. But whether the solution is Ballista or another piece of software, Ernst's UCS Architecture, if successful, has the potential to dramatically change robotic warfare.

First, there are tactical implications for single-drone controllers. Units tend to fly one robot at a time. While that's often a function of drones' limited availability, especially for small units, the inability of controllers to toggle from one drone to another doesn't help. Individuals who remotely pilot one system can't necessarily transfer those skills to another. That keeps the military services buying redundant robots, instead of consolidating control, and flying them solo.

It also means the robots can't talk to one another. The information their sensors collect is transmitted ' apparently without encryption ' to distinct ground control stations. With a common drone OS, all that data ' full-motion video of a suspected insurgent hideout; a stretch of road known for a high concentration of insurgent bombs ' could show up on a single screen.

And that sets up perhaps the most dramatic change of all. With a common controller, remote pilots can put together fleets of drones, including multiple types of drones, for a single mission. Admittedly, that's not just a software problem, it's a hardware problem: not many military commands control, say, a robotic helicopter scout and an armed Predator. But if there's a common drone OS, then it's possible in theory for a unit to use both at once for different aspects of the same mission, and control them all from the same machine. Picture a commander ordering one of the Air Force's forthcoming bird-shaped micro-spies to scout an urban area for an enemy while keeping a Predator loitering 30,000 feet overhead, waiting for the robo-bird to find its prey.

All this represents a very big technological step, with accordingly large implications for the future of drone warfare. Ernst doesn't say how long it'll take. But if his office can pull it off, the military won't just get all its drones run by a standardized piece of software. It may get the ability to control whole fleets of deadly flying robots at once.