Like all branches of the U.S. military, the Marine Corps has invested heavily in unmanned systems in recent years. When the Marines spearheaded the drive through southern Iraq six years ago, they owned just a handful of 20-year-old Pioneer Unmanned Aerial Vehicles. Today, the Marines operate newer Shadow UAVs at the division level, Raven UAVs down with the infantry battalions and companies, and small numbers of Packbot and Talon ground robots for engineer and bomb-disposal teams. While widespread, Marine robots fill mostly niche roles.
Now the officer that oversees Marine robot development wants bots capable of truly fighting alongside Marine riflemen, in large numbers. “You need to help me give them a better robot to operate more like a member of the squad,” Colonel James Braden told an audience of engineers and managers at the annual Unmanned Systems conference in Washington, D.C., last week.
Squad bots would perform a wide range of missions. Some, called “mules,” would haul supplies for the over-burdened infantry. Others would carry weapons to complement human fire teams. Both types of bots already exist in prototype form, but require significant improvement before they can “enlist” for day-to-day service.
The mules could be based on Boston Dynamics’ Big Dog bot, which looks like a headless horse and can “prance” over rough terrain while carrying loads of hundreds of pounds. Or, the mules might be wheeled, like those that Lockheed Martin was developing for the Army’s now-canceled Future Combat Systems. Legs would work better on rough terrain; on roads, wheels out-perform legs.
Regardless of the form the cargo bot takes, it must be adept at following the infantry, Braden said, adding that “it’d be nice if it didn’t make a whole bunch of noise.” The defense industry has experimented with a wide range of “following” systems for robots. Most rely on an optical or laser sensor that can identify the “leader” and keep the bot moving towards him. These bots also need some way to sense and avoid obstacles. Braden said today’s robots tend to get confused when they encounter people they don’t expect — especially short people and children, who might not match the bot’s standard definition of a person. Puddles and other flat obstacles also trip up today’s mule-type bots.
The search for a robotic mule reflects the Marines’ growing emphasis on unmanned logistics. This spring, Marine General John Amos told Congress the Corps would survey industry for a UAV capable of quickly transporting nearly a ton of supplies to isolated, front-line forces. U.S. Special Forces already use the NMIST Snow Goose bot, deployed by parachute, for cargo delivery. The U.S. Army does not share the Marines’ enthusiasm for robotic resupply. “I’m not sure it’s cost-effective,” said Colonel Gregory Gonzalez.
The Army is also skeptical of armed robots. In addition to adding cargo mules, the Marines want to eventually reinforce infantry platoons with robotic riflemen. The Army tried this with its prototype SWORDS robots in Iraq in 2007. The machine gun-armed, tracked robot (pictured), built by Foster-Miller, never saw combat. The Army, not trusting SWORDS’ reliability, kept the bot confined to base.
To work in combat, an infantry bot will need to have the same “muscle-memory” that human soldiers do, Braden said. The bot must be able to sense and respond to threats, very fast, while ignoring the many distractions of the chaotic, urban battlefield. Contrast that to the slow, methodical processes the Air Force uses when firing missiles from its Predator and Reaper UAVs.
“In the air, we have a very disciplined and regimented way to do engagements,” Braden said. “In the ground fight, it’s much more of a snap fight.” With firefights lasting just seconds, there’s no time for a human operator to micro-manage a robot’s reaction. The bot will have to detect and orient towards threats, all on its own, using some sophisticated blend of optical and audio sensors, and perhaps even radar. The robot would then ask for permission to open fire. “There is always a man in the loop, pulling the trigger,” Braden said.
Braden said the Marines will begin testing some of the basic processes that might result in an infantry bot, in the “next year or two.” Some of the biggest obstacles are conceptual. “One of real challenges in ground bots is building confidence in our tests for what autonomy means.” In other words, what should we realistically expect from a robot, and what should we resign to doing ourselves — at least for now?
(Photo: via Popular Mechanics)