Killer Robots - Army Studies Challenges of Remote Lethality

The military has used and experimented with robots that perform functions such as scouting and surveillance, carrying supplies and detecting and disposing of improvised homemade bombs. However, when it comes to integrating lethality, such as a weapon capable of firing 10 rounds per second onto an unmanned ground vehicle, issues arise such as safety, effectiveness and reliability, as well as military doctrine on how much human involvement is required.

Advanced Remote/Robotic Armament System, 7.62mm Prototype. (Photo: U.S. Army)

Robert Testa, the technical lead of the Remote Weapons Branch at Picatinny Arsenal, New Jersey, recognizes the growing evolution in autonomous technologies and is focused on improving existing remote weapon technologies for manned and unmanned platforms, as well as fixed‑site applications. Testa, whose branch is part of the Armament Research, Development and Engineering Center (ARDEC), said the term supervised autonomy strikes a contemporary balance between schools of thought that range between total human control (tele‑operation) and researchers who are developing the technologies to enable robots to think for themselves. Tele‑operated means a human makes all the decisions regarding the activities of a remote platform, which is linked to the operator through a radio frequency or a physical link such as copper cable or fiber.

Regarding the term supervised autonomy, Testa said, "I believe that UGV (unmanned ground vehicle) and robotic platform developers apply the term supervised autonomy because not only do robust fully autonomous ground platforms still require substantial development, but it is essential that any UGV have the capability to react to command and control from a human operator under certain circumstances. This is similar to the addition of limited autonomy to RWS (remote weapon systems), but our primary focus has to be the robust, real time, tele‑operation capability to ensure safe and effective weapon operation, regardless of the platform or application."

The term "supervised autonomy" also reflects the current state of technology. Testa favors the term unmanned ground vehicle, which can encompass either a tele‑operated platform or a robot with varying degrees of autonomy, as a way to distill the mission of the remote weapons branch at Picatinny with respect to remote lethality.

"There are some areas where we and our partners are developing the capability to add degrees of supervised autonomy for weapon systems, but we recognize that you will always require real‑time manned supervision of what the RWS is aiming at, what targets are engaged, and when that engagement takes place, i.e, trigger pull," Testa said.

Army research into remote lethality complies with Department of Defense Directive 3000.09 "Autonomy in Weapon Systems," published in November 2012. The protocols reflect the current doctrine addressing all classes of remote and unmanned weapon operation designed to allow commanders and operators to exercise appropriate levels of human judgment over the use of force.

"We can easily enable current generation remote weapon systems to aim and engage targets autonomously," explained Testa. "Yet current doctrine and the realization that current sensor and processor technologies would provide little or no assurance to what was engaged, keeps the real time tele- operation mode of weapons use at the center of what we develop and demonstrate today. Today's current remote weapons systems are primarily developed and deployed as tele-op weapon systems, yet they inherently lend themselves as the starting point for the future of UGV lethality. The [Remote Weapons Branch] is working to make remote weapons more remote."

Researchers are aware that the term remote weapon may evoke an image of something operating many miles away, with a high degree of autonomy; however, remote could also mean a weapons system on top of a vehicle with the operator inside under the protection of armor. At ARDEC, a remote weapon system is closely associated with something like the fielded Common Remotely Operated Weapon Station, also known as CROWS. CROWS is a stabilized mount that contains a sensor suite and fire-control software. It allows on‑the‑move target acquisition and first‑burst target engagement. Capable of target engagement under day and night conditions, the CROWS sensor suite includes a daytime video camera, thermal camera and laser rangefinder.

CROWS supports the MK19 Grenade Machine Gun, the M2 .50 Caliber Machine Gun, M240B Machine Gun and M249 Squad Automatic Weapon, weapons originally designed for manned operation. The system has been integrated onto more than 20 platforms, from the Humvee to the M1 Abrams tank. Yet current‑generation remote‑weapons systems as CROWS cannot support other functions essential to making remote weapons more remote.

The ARDEC‑developed Advanced Remote Armament System (ARAS) has additional capabilities, such as an externally powered, purpose‑built weapon to improve reliability and accuracy, the ability to load and clear the weapon remotely and an increased stowed ammunition load without decreasing aim or stabilization. It can also reload the weapon or change ammunition type without manned intervention at the weapon, in approximately six seconds. Also critical for future asymmetric engagements is the ability to change from lethal to non‑lethal ammunition that ARAS provides. ARDEC has developed both 7.62mm and .50 caliber ARAS prototypes. The ARAS patents are owned by the U.S. Army, which enable cost‑effective acquisition once future requirements are generated.

Although ARDEC does not develop vehicles or platforms for its weapons, it has used a tele‑operated MS3 Ripsaw as a "surrogate platform" for the development of UGV lethality technologies, including wireless extension kits for CROWS, ARAS and other remote weapons systems. These programs have culminated with the first unmanned ground vehicle Scout Gunnery Table VI experiment in November 2013, at Fort Dix, New Jersey. ARDEC also works with other government and industry partners to arm UGV platforms.

While some UGV functions may include such things as carrying equipment, surveillance or removing homemade bombs, their functions are not lethal in nature. But incorporating powerful weapons onto a UGV presents a number of technical challenges, including minimizing or eliminating latency. Latency means delay or the time elapsed between using controls to initiate an action and when it actually happens. This is particularly critical when you have high‑rate‑of‑fire weapons, Testa said. Latency also applies to video, or the time elapsed between the images captured by a RWS mounted camera and when they actually appear on the screen of the operator.

"Latencies are bad, and they are technically challenging, especially video latency," Testa said. "Our goal is to minimize video latencies as much as possible and we are targeting a maximum of 250 milliseconds, or quarter of a second."

Testa said humans start to notice latency at around 250 milliseconds. "You start to sense, 'I moved the joystick, but I didn't see the reticle move right away.' You want your video to be real‑time as much as your controls are real‑time. I pull the trigger, I want the gun to shoot immediately or close to it as possible, but I also want to know what I'm looking at as close to real time as possible."

The lethal nature of a weaponized unmanned ground vehicle, and the need to keep latency to a minimum, are reasons that a robust connection from the operator to the remote weapon system is critical, whether that link is by radio, fiber or copper. Testa said that when people hear about predator drones hitting their targets while the operator is halfway around the world, that may leave the impression that something similar can occur with remote weapons systems on the ground.

"They are using satellite communications with a fairly large latency," Testa noted of such drones. "We're shooting machine guns in a cluttered and asymmetric ground environment. That makes the problem a lot tougher in some ways. Drones are targeting for and launching one missile that has guidance to the threat. We are shooting dumb bullets at 10 per second, so we don't have a lot of wiggle room."

Source: