Budgets Permitting, Marines Could Be Fighting Alongside Robots by 2020s
With fiscal upheaval in the Pentagon, it is uncertain how much funding will be available to purchase robotic systems, but industry and military laboratories are working on the Marine Corps’ behalf to develop technologies and drive costs down, said Roy Byrd, director of government relations and Marine Corps programs at ITT Exelis. Because of the budget crunch, labs like the Office of Naval Research are focused on retrofitting existing vehicles and systems with autonomy kits that will allow them to operate without a human in the driver seat.
“The science and technology focus is developing autonomy enablers, not platforms,” Byrd said. “The [science and technology] challenge, like for everyone else, is affordability. The service cannot afford, nor is it reasonable, to spend hundreds of thousands of dollars more on sensors.”
Byrd spoke on behalf of ONR at the National Defense Industrial Association’s annual Expeditionary Warfare Conference in Portsmouth, Va. Robots will be able to work and fight alongside Marines within 10 years, he said. Autonomous systems will first be developed to act as force multipliers by carrying supplies, evacuating casualties and eventually will create new concepts of operations for ground troops, he added. But their widespread introduction is heavily dependent on affordability, given constrained Marine Corps spending.
“Aside from possibly a few niche applications, autonomous ground vehicles are unlikely to be effectively integrated into the operational force until implementation costs come down,” Byrd said. “Marines cannot afford to have an autonomy package that costs substantially more than the base vehicle platform itself.”
ONR plans to develop robots that can autonomously connect Marines in the field with supplies and medical evacuation. That includes driverless point-to-point navigation for routine and emergency resupply, and casualty evacuation for small units ashore. ONR is developing, in concert with industry, an autonomy retrofit kit that can be easily and affordably installed in the service’s existing vehicles. There will be a demonstrator for these capabilities on an internally transportable vehicle — which fits into a V-22 Osprey — by fiscal year 2016, Byrd said. Ideally the kit will cost under $25,000.
Lockheed Martin successfully retrofitted two K-MAX helicopters with autonomous flight computers. The aircraft have delivered millions of pounds of cargo to Marines at forward operating bases in Afghanistan.
Troops at the Marine Corps Warfighting Laboratory have extensively tested two vehicles that can operate without human drivers by way of add-on autonomy kits. The Cargo Unmanned Ground Vehicle, an autonomous 7-ton truck developed by Oshkosh Defense, is designed to reduce the number of troops needed to run supply convoys — a frequent target of insurgent ambushes and roadside bombs.
Using the autonomous drive system, Marines can reduce the personnel required for a typical convoy by 19 percent. An eight-truck resupply mission, for instance, could be carried out by 26 troops instead of 32. As the technology matures, that number likely will reduce. The trucks can be set to continuously run designated routes with complete autonomy or be remotely driven by someone at a ground station or riding in another truck. Because the autonomous drive system is retrofitted onto existing vehicles, with the flip of a switch they can be driven manually. But the trucks only reduce the number of Marines needed to transport supply rather than eliminating them from the cab.
“Current technology is largely based on remote-control systems with humans always in the decision loop, largely limited to line-of-sight communications and dedicated single-purpose platforms,” Byrd said. “ONR’s objective is to bring to the Marine Corps technology that will allow increasing levels of autonomy … relieving the human of physical and cognitive burdens.”
Gen. John Paxton Jr., assistant commandant of the Marine Corps, said technology has made small units more capable, but cannot solve all the problems Marines face in combat. He warned in October of becoming too reliant on technologies that could potentially cause, rather than alleviate, logistics challenges from ship to shore and between bases and Marines in the field.
“Now we are doing things with a platoon that 10 years ago you would have to go to a brigade to do in terms of maneuver and multiple lines of communication and things,” Paxton said in October at the Association of the United States Army’s annual conference in Washington, D.C. “The corollary is you now have power problems, and logistics sustainment problems, lift problems that are endemic to a platoon and company size units that you would used to not even worry about.”
Lt. Gen. Richard P. Mills, Marine Corps deputy commandant for combat development and integration, witnessed Marines use some semi-autonomous vehicles during field exercises in 2012. Mills told National Defense then that he is open to any technology that would improve efficiency and keep troops safe.
The systems must be thoroughly tested before being bought in large numbers. The same fiscal constraint that makes autonomy attractive compels the services to vet them heavily and choose wisely, he said.
“Our decision is going to be tough because we don’t have a lot of money. So we have to make sure these things are well thought through,” Mills said. “Sometimes failure is just as valuable as success. We may find out it’s a nice thing to have but it doesn’t really work on the battlefield.”
Marines have also trained with the ground unmanned support surrogate, a four-wheeled gas-powered offroad vehicle retrofitted with a robotics kit that allows it to automatically follow a Marine at a specified distance or navigate routes autonomously from point to point. Smaller and more agile than a 7-ton truck but lightly armored, the GUSS, developed by Blacksburg, Va.-based TORC Robotics can evacuate casualties from combat situations, do small-unit resupply and carry gear for a squad.
The Defense Advanced Research Projects Agency is nearing final testing of a robotic mule that is purpose built to carry supplies for Marines on the move.
The Legged Squad Support System, or LS3, represents the latest evolution of a technology DARPA has poured $54 million into since 2009.
DARPA’s goal was to build a robot that mimics the load-bearing abilities of an actual mule, so troops could pile their gear onto it while traveling. More specifically, they aimed to create a robotic platform that could carry multiple 40-pound packs and walk at least 29 miles in a day without intervention. The mule eventually will provide auxiliary power to troops in the field, reducing their reliance on bulky batteries.
The LS3 will undergo a two-week test in March at Marine Corps Air-Ground Combat Center 29 Palms, Calif. Other tests are scheduled at the Marine Corps Mountain Warfare Training Center in Bridgeport, Calif., and at Ft. Benning, Ga. Final tests are tentatively set to coincide with the 2014 Rim of the Pacific exercises.
The robotics development timeline calls for LS3-style autonomous ground vehicles acting as “wingmen” sometime before 2022. That will require advanced machine intelligence that behaves in a reasonable, appropriate and transparent way that Marines understand, Byrd said. The system must interact with human teammates in natural ways such as speech, gestures and physical contact, he added.
“Autonomous machines [will be] serving as an additional squad member demonstrating independent tactical maneuver and actions,” Byrd said. “Robots with advanced intelligence will operate with an infantry squad behaving much as another human.”
That means robots should recognize and adapt to tactical situations and threats the same way a human would. Not only should the robots be able to take cover when fired upon, but the system should instinctively place itself between incoming fire and human troops, Byrd said.
Beyond 2023, the probable capabilities and missions of autonomous systems become less clearly defined, Byrd said. ONR is seeking what it calls “integrated multi-role systems” that can think on the fly, interact seamlessly with troops and possibly carry weapons.
“Advanced learning and reasoning skills will allow complex tactical behaviors,” he said. “Systems may be weaponized with no human in the loop.”
Achieving that level of performance requires advancements in sensor technology and computing algorithms that detect threats and direct a system’s behavior, said Craig Woolsey, an associate professor of aerospace and ocean engineering at Virginia Polytechnic Institute and State University.
“It’s not about the vehicle, it’s about the sensors and the algorithms,” Woolsey said. “There’s no such thing as an autonomous vehicle. They are built by humans to help people. … The person is either a passenger or a supervisor or maybe an antagonist.”
Woolsey is focused on developing software to improve environmental perception of sensors mounted on vehicles. Rather than machines simply “seeing” an environment, they need to understand how certain factors can affect their capabilities and be able to predict changes as a human would, he said.
“Our focus is perception and cognition for autonomous vehicles operating and increasingly cooperating in dynamic and uncertain environments,” Woolsey said. “Not only are the environmental characteristics changing — the weather, the conditions, possibly traffic that you have to deal with — but there might also be antagonists. That requires real-time adaptive behavior and human-machine interaction.”
The latter is a difficult cultural and technological hurdle to overcome because Marines must learn to trust that robots will act in the troops best interest at the appropriate time, he said. Autonomous systems must understand that snow or rain could hinder their ability to move through an environment, for example, and must follow a small-unit leader’s orders without him having to modify the way he would communicate with any subordinate human, Woolsey added.
DARPA is studying many of those technological challenges through its Robotics Challenge that began in October 2012. The competition is asking engineers from industry and academia to develop ground robots capable of executing complex tasks in “dangerous, degraded, human-engineered environments and use tools and equipment ranging from hand tools to vehicles.”
The program is scheduled to run 27 months with three individual competitions. A virtual competition was held in June. Real-world trials began in December and will conclude with another series of competitive tests at the end of 2014.