Army Floats Plan for Autonomous 'Wingmen'

Maj. Mike Dvorak, robotics branch chief at the Army Capabilities Integration Center, wanted to make a point right off the bat during a recent talk. 

“There is no plan to have autonomous killer robots, or anything like that,” he said at the National Defense Industrial Association’s Robotics Capabilities Conference in March.

Just in case anyone missed it, he reiterated the statement later.

Nor is the Army intending to use machines to replace personnel, he added. The service on Feb. 27 published its autonomous systems and robotics strategy. Dvorak, who led the study, came to the conference to discuss that, and the service’s “tentative” roadmap to field robotic wingmen for its tactical wheeled and combat vehicles.

While the idea has been discussed for years, it began to gain purchase among senior Army leaders last September, Dvorak said. He has been working on a draft plan since then. 

The “wingman” term — with its roots in the world of military aviation — is spreading beyond the skies and has been discussed for several years as a concept the Army might employ. It would mean battlefields populated by a mix of manned and unmanned vehicles, some of them armed, although Dvorak also wanted to stress that a human will remain the ultimate decision-maker as to whether lethal force is employed — as per Defense Department policy. 

“People see robotic wingman and they think robotic tank. And I’m going to say again, let’s be clear that what we’re really doing is focusing on mobility,” he said.

Autonomy is still important, although not for employing lethal force, he said. For example, an unmanned vehicle must be able to do obstacle avoidance without needing a soldier to operate it. The idea is to free up troops to do other tasks — not create more work for them.

“We want that vehicle to stop, slow down and do some things on its own,” he said.

Remote weapon systems may be mounted on the vehicles. That’s not new, he pointed out. The common remotely operated weapon station (CROWS) allows soldiers to shoot at targets while bottled up inside their fighting vehicles. It has been around for more than a decade.

However, the gas-powered systems are analog, not digital. If they jam, that poses a problem because no one will be aboard to fix it, he said.

Latency issues will also have to be worked out. A Hellfire missile — once launched — can continue to track a target even if there is a 30-second delay in communications. Machine gun bullets can’t do that. Dvorak said once a target is chosen by the human operator, sensors will have to take over to ensure the gun keeps track of the target, suggesting some autonomy in the weapon system.

“For a machine gun, we don’t have smart bullets, so sensors are going to have to do the job,” he said.

And as far as sensors go, the robotic wingman will have to match the capabilities of a tank platoon. There are four tanks, and four operators within each, all of whom are keeping an eye out for trouble. One sensor placed on the front of the vehicle won’t cut it, he said.  

“What about the other 359 degrees?” he asked.

Speed is also a problem. Tests so far have shown Humvees rigged to drive themselves losing control at about the 25-miles-per-hour mark. Future vehicles will at least have to go as fast as a tank. 

One of the biggest challenges will be networking all these manned and unmanned vehicles and their sensors, he noted. How are they communicating? How are sensors talking to other vehicles and how will an unmanned vehicle send targeting data to a manned vehicle and how will all that touch the network? he asked. 

These are some of the questions and problems the Army hopes to address this summer at an exercise at Fort Benning, Georgia. ARCIC is seeking approval to do a joint capability technology demonstration.

One of the new technologies that will be shown there is an automated munitions loader for tank crews. If that is successful, it would presumably free up a crew member to be the trigger puller for a robotic wingman, he noted.

For now, ARCIC is leaving robotic tanks alone, although it is relatively easy to convert about any vehicle to remote operations with a kit, he said. In the initial stages, the draft plan calls for robotic wingman Humvees and M113 armored personnel carriers. Army leaders are particularly interested in using a robotic vehicle to do smoke obscuration. The tactic, which is the responsibility of the Chemical Corps, requires personnel to drive between an attacking force and a defending force with a smoke generator attached.

“It’s probably one of the most dangerous tasks that the Army can do,” he said.

The August demonstration would also involve a Humvee outfitted with a remotely controlled machine gun, as well as small unmanned aerial vehicles to do targeting and battlefield damage assessment, he said.

The tentative plan calls for converted Humvees and M113s in the next few years, then a purpose-built robotic wingman five to 10 years later, he said. First, the Army wants to produce a robotic wingman development document by Oct. 1, 2018, he said.

“Right now, we are payload and platform agnostic. We don’t know exactly where it is going to go,” he said.

Despite Dvorak assuring the audience at the conference no less than three times that autonomous armed robots weren’t on the table, former Undersecretary for Acquisition, Technology and Logistics Frank Kendall wrapped up the two-day show with a speech suggesting otherwise. 

“Lethal autonomy is going to happen,” he said. “We are well down the road ... of having effective lethal autonomous systems.”

Topics: Robotics, Autonomous, Robotics and Autonomous Systems, Science and Engineering Technology