The US Army aims to take many more soldiers off the battlefield - but these drone convoys bring new challenges
- The Army is preparing to test driverless vehicles in the convoys that deliver supplies like food and ammunition to bases and outposts.
- The service sees autonomous convoys as a way to free up troops and lower the risks they face.
- While introducing autonomous tech is feasible, doing so will raise more challenges, requiring more changes going forward.
The US Army wants to free up manpower and take troops out of harm's way, and one part of that push are self-driving convoys. Think autonomous Teslas, but with five-ton trucks that keep front-line troops supplied in war zones.
This kind of technology is already available and getting better every day, but introducing it to combat is only the next step, and the Pentagon will soon find itself dealing with new tactics deployed against their driverless vehicles.
Army Secretary Mark Esper said in May autonomy is one of the service's "key technologies ... critical to the Next Generation Combat Vehicle," which is one of the Army's big six modernization priorities, meant to replace Bradley fighting vehicles and eventually Abrams tanks with a combination of robotic and manned vehicles.
The push is not limited to combat vehicles.
The Army has been looking at self-driving resupply vehicles for years under the Autonomous Ground Resupply program. As part of that effort, the service is not gearing up for the Expedient Leader-Follower demonstration, or ExLF, according to an August report by Breaking Defense.
Maryland-based firm Robotic Research was awarded a three-year, $50 million contract this summer to provide autonomy kits for Army vehicles for the demonstration, with the goal of "showcasing the integration of modular kits, common interfaces, and a scalable open architecture," the firm said in a release.
Contractors are already putting the kits onto 10 Oshkosh palletized loader system trucks for safety-certification tests in 2019. The Army wants 60 more trucks equipped to be self-driving by 2020 for a large-scale demonstration.
The Army is pursuing ExLF because it still needs a human involved - one to drive the lead vehicle, guiding other self-driving vehicles that may still have humans aboard to take over if there's a problem or to respond to external threats.
"This follow-the-leader style, that's not a big technology leap," said Peter Singer, a senior fellow at the New America Foundation and a specialist on 21st-century warfare. "Frankly you can see a lot of that technology already woven into the cars that are coming out in 2019, 2020" for sale to the public.
The idea of taking troops out of the resupply role has been bouncing around the Pentagon for some time.
In early and mid-2000s - when a large percentage of US casualties in Iraq and Afghanistan were sustained by improvised explosive devices and ambushes during convoy operations - military researchers wanted to move forward with driverless-car technology, Singer said.
Contests held by the Defense Advanced Research Projects Agency around that time featured robotic cars developed in the private sector racing over "terrain was uncoincidentally like the terrain you would have to navigate in Iraq and Afghanistan," Singer said. "This idea of doing it to save lives, to save personnel, there's a long history to it, and it makes a great deal of sense."
The private sector has largely taken the lead on self-driving technology, but there has been renewed interest in the Pentagon, particularly among Army leaders, who want to field totally unmanned convoys in a later stage of the Autonomous Ground Resupply program - perhaps as soon as 2022.
Oshkosh and Robotic Research told Breaking Defense that could be done sooner, but the Army is not yet ready, and fielding vehicles without human drivers is likely to raise new issues.
US Army/Staff Sgt. Kathleen V. PolancoGermans watch US Army vehicles drive on German roads, April 22, 2018.
Setting a route and getting an unmanned vehicle to follow it is possible with current technology, but they may encounter problems that robots and computers aren't yet equipped to handle.
Self-driving and semi-autonomous cars fielded by companies like Uber and Tesla have misidentified objects on the road and misjudged other vehicles, sometimes with fatal results. Similar problems could arise in combat.
"A little kid runs out into the street, and a convoy runs them over, and it's not just a tragedy. It maybe loses the hearts and minds of that village," Singer said. "Maybe it sparks protests or even insurgent violence."
Programming autonomous vehicles to stop when they encounter a person or object in the road may create new vulnerabilities, as a savvy adversary could deliberately block roads to stop convoys.
"War's always about a back and forth," Singer said. "If the machine is programmed every time a civilian's in the roadway [to] stop, well, you might have an adversary go, 'OK, cool, we'll take advantage of that. We'll have humans go into the road.'"
The links between self-driving resupply vehicles and the soldiers responsible for them would also be vulnerable.
US Army/YouTubeArmy vehicles taking part in testing for the Autonomous Ground Resupply program.
"The challenges to autonomous vehicles ... are the challenges of the nonlinear battlefield," said George Topic, vice director of the Center for Joint and Strategic Logistics at the National Defense University.
They are "challenges of information security, i.e. that you may be able to produce significant damage to us just by ... interfering with how we communicate with autonomous vehicles."
An autonomous resupply vehicle could be told to travel between "these GPS waypoints and no human interaction [is] needed," Singer said. "OK, well what happens when GPS goes down because there's hacking or electronic warfare?" (Other guidance systems are available, but geography and adversaries can still hinder them.)
The US military is already preparing for electronic warfare to play a larger role on the battlefield. Mitigating that challenge raises more questions about what autonomous vehicles will need to operate effectively.
It may mean adjusting the technology. It may mean new technology is needed. It may mean humans still need to be involved.
One solution may be "just keep the humans in there and just have less of them. It'll be kind of a team," Singer said. "Another may be a drop-in model, where the machines are operating autonomously, but if something interesting ... happens, the human can drop in remotely and control it."
An alternative, like cargo-carrying drones, may be better in some situations, Singer said. "Another alternative is ... just make it smarter. This is why we need the machine to be smarter and better."
"It can't just blindly follow a route," he added. "It's got to be able to react to all these circumstances."
Thomson ReutersUS Army trucks get a thumbs-up from a fellow soldier after crossing into Kuwait during the last convoy out of Iraq, December 18, 2011.
"We're going to need to think about the second- and third-order effects of this as well," Topic said.
"Autonomous trucks are going to be equipped with really robust diagnostic sensors, but when you're driving a car and you hear it go 'clunk, clunk, clunk,' or something doesn't work the way you want, you say, 'hmmm, something's wrong,'" he added. "In the same way with these vehicles ... we need to figure out how to deal with those kinds of things."
The problems the Autonomous Ground Resupply program will face in the future are not yet known, and the final form of the program is yet to be seen.
"The technology aspect of this stuff I think is all very plausible. It's the question of how you use it," Topic said. "But I think these are very important ideas, and I think that moving faster on this stuff is certainly worthwhile."
"It's not a significant investment for 70 trucks when you consider the kind of money we spend overall," Topic added. "The most important factor is the security of the people we are putting in harm's way."
The program should be seen as a first step, according to Singer, who said the timeline and small-batch buy were wise.
"That's frankly contrary to a lot of other military programs where a big mistake ... has been to pre-commit to a massive buy of a certain model before you know not just whether it works but how it works," he said.
"This is a better example of, get a small number of them, test them, learn from them, put them through the paces, and then do it again and again and again."