Army Wants Robot Four-Wheelers and Decepticons

Nearly 10 years ago, the Defense Advanced Research Project Agency, or DARPA, showed that a self-driving car could navigate a 132-mile stretch of desert. That proof-of-concept test, DARPA Grand Challenge (2005), sought simply to illustrate how military resupply might be made less risky. (It also became the basis for Google's self-driving car program.)

What does the Army want from robots 10 years from now? In short, to make human soldiers more lethal in combat, make the job of soldiering less deadly, and turn dismounted patrols into a high-performing mix of soldiers and semi-autonomous machines. That includes programs that haven't been formally announced, such as one called Mobile Infantry, also from DARPA. (You read it here first.) It will mix soldiers with, essentially, self-driving all-terrain vehicles during combat operations.

Lt. Gen. H.R. McMaster, deputy commander of the Training and Doctrine Command, or TRADOC, has commissioned a doctrine on autonomy and robotic systems. Expected for release in January, it will be a follow-up to November's Army Operating Concept for 2020-2040 and will lay out the Army's expectations for robotics.

But its outlines are already visible in the Operating Concept, which says, "The application of emerging technology creates the potential for affordable, interoperable, autonomous, and semi-autonomous systems that improve the effectiveness of soldiers and units. Autonomy-enabled systems will deploy as force multipliers at all echelons from the squad to the brigade combat teams."

For months, military leaders eagerly have emphasized robotic autonomy as a key component of future science and technology development. But not everyone is looking to replace every human in every situation with a bunch of metal and wires. Ten years after the onset of self-driving cars, military officials are still asking themselves how smart, or autonomous, they want robots to be and what they want robots to do.

The doctrine will be a "clear articulation," of the Army's near-, mid-, and far-future expectations and desires for robots, Army Lt. Col. Matt Dooley recently told an audience at the National Defense Industrial Association in Virginia. "Autonomy-enabled systems are the next revolutionary capability for force 2025 and beyond. The Army envisions a future operational concept where autonomy-enabled formations augment the war-fighter as team members, not just as tools."

Here are some of the things the Army wants from its future bots, according to Dooley:

1. Allow forces to maneuver out of contact and exploit known enemy positions. This requirement has to do with the military's efforts to avoid the homemade bombs known as improvised explosive devices, or IEDs. The military recently all but shuttered its Joint IED Defeat Organization, or JIEDDO, but "that mission is not going away," said Dooley.

2. Conduct logistical resupply. Robot resupply programs aren't new, but the need for them is growing. As the Army loads more gear on its patrolling soldiers, knee and back injuries have risen quickly, according to Brad Tousley, who directs DARPA's office of Tactical Technology.

"One of the things we're very concerned about is reducing that burden," said Tousley, "treating the squad as a system … not giving everyone in the squad the same thing [heavy piece of equipment]. Using robotics on the squad level can add to that."

That imperative led to BigDog, the famously loud four-legged robot pack animal from Boston Dynamics. The company emerged from the DARPA's Legged Squad Support System program, or LS3, and was acquired last year by Google.

The military envisions squads of dismounted Marines moving through hills or cities, accompanied by BigDogs carrying 50-pound machine guns, food, and water. Increasing the level of autonomy is crucial: a dangerous urban battle space is a terrible location to have your eyes buried in some sort of steering app.

For a sense of what that integration will look like, check out DARPA's Squad X Core Technologies program. Announced last year, it aims to allow dismounted units to use multiple walking, flying, and other drones. The objective is to get robots in formation and have them behave much more like well-trained animals than machines in need of constant supervision.

This is where the "Mobile Infantry" program comes in. The combination of soldiers and self-driving ATV-like machines will "allow for a combined set of mounted and dismounted operations and for a larger area of operations over more aggressive time lines than standard infantry units," according to a description on one of Tousley's slides. Today, Mobile Infantry is what Tousley called a "proposed program," one that hasn't yet been approved by DARPA's director.

The military is developing autonomous trucks that can follow each other in a convoy. See what that looks like in this video of Army trucks that drive themselves.

3. Set the conditions for manned force entry. In other words, be the first thing in the room and set the stage for the soldiers to enter. That could mean everything from turning off the lights to providing real-time video feeds of the room's layout and other occupants.

4. Enable smaller manned expeditionary force advantage. That's another way to say robots should provide situational awareness and load-bearing capabilities so soldiers can do more.

5. Present multiple dilemmas to enemy and enhance manned freedom to maneuver. To pull this off, the robots would undertake "deception operations." The Army doesn't go into much detail as to what those are, but it could include things like decoy drones that land in a room and make noises that draw enemy fire and drones that blow smoke creating the impression of an bomb blast to frighten enemies away.

6. Enhance security for manned units and increase the depth of knowledge for the enemy,or what the Army calls "persistent surveillance." One form that could take is robots that are left in key settings for extended periods of time—possibly years—and that become active when they sense a threat or change. Imagine someone enters a room with an object that the robot's vision system understands to be a gun. It's a concept not dissimilar to DARPA's Upward Falling Payloads Program, which Defense One politely rebranded "zombie pods of the deep."

7. Conduct land operations in a tactical chemical, biological, radiological, and nuclear warfare environment to "mitigate risk and allow freedom of maneuver of manned forces." That's the focus of the DARPA Grand Robotics Challenge, which was modeled loosely on the environment of the Fukushima Daiichi Nuclear Power Plant circa March 2011.

In addition to the above, Dooley also outlined two specific items that the Army is definitely not pursuing. One is in losing capabilities they already have. Steering robots is a chore but it's one the Army wants to preserve the right to carry out. No matter how advanced machine intelligence becomes, the Army wants to hold on to and strengthen the ability to operate machines remotely.

"There's been a lot of great work in teleoperation. Just because we're pursuing autonomy doesn't mean we want to let that stuff go. We want direct human intervention," at times, said Dooley.

The other item the Army says it won't pursue are killbots. Today, a 2012 policy directive prohibits the military from creating robots that can "select and engage individual targets or specific target groups that have not been previously selected by an authorized human operator." Dooley says there's no effort underway to change that directive.