The Pentagon’s funding of MIT’s “beerbots” is getting some attention lately. Congress, reasonably, has posed the question of, “Why is the Pentagon researching beer delivery robots, especially while hotels and bars are already deploying robot bartenders?”
Well, the answer is a little more logical than you might think. So, Alexa, crack open a cold one and let’s talk about beerbots.
Waiters that are part of MIT’s “beerbot” program go into an office to work with humans.
First off, we think it’s awesome that Congress accepted the possibility that the military was researching beer-delivery robots in order to distribute cold beers more cheaply (and was seemingly okay with it so long as it wasn’t redundant). That being said, the actual MIT program is focused on figuring out how to get robots to best coordinate their actions in uncertain environments, something that could prove vital for everything from future hospitals to underground fighting.
See, MIT was building a system of cooperative robots, robots which could communicate with each other and share sensor data and other observations to work more efficiently. When they designed a complex, real-world situation to test them in, one obvious angle was to have them serve drinks in an office. And, surprise, the drink that graduates students want is beer.
And so, the “beerbots” were born. There’s a “PR2” robot that picks up drinks and places them in coolers which are carried by the “turtle bots,” and the turtle bots act as waiters. The turtle bots move from room to room, taking orders and either filling the orders or marking that the room has no orders.
And here’s the key part: The robots share their data with each other. The PR2 doesn’t know what orders are placed until the turtles get close, and the turtles rely on each other to map out routes and obstacles and to share drink orders to figure out the most efficient path to fill them.
Soldiers with the 1st Battalion, 6th Infantry Regiment, take part in an Army Asymmetric Warfare Group program designed to improve military tactics, techniques, and procedures while fighting underground.
(U.S. Army photo by Lt. Col. Sonise Lumbaca)
This is actually a complex logic problem for the bots when they also have to deal with humans moving from room to room and constantly creating and changing obstacles in the office.
And this is basically the starter level for robots that could help humans on battlefields of the future. Take subterranean warfare, an area so important that the U.S. is considering naming it as a new warfighting domain, for example. Robots helping humans underground will be physically limited in how they can communicate with one another as concrete or subterranean rocks block electromagnetic signals and lasers. So, robots will need to aid the humans there by carrying loads or ferrying supplies, and then communicate directly with one another to determine what’s going on in each section of the underground network.
Paratroopers with 1st Battalion, 508th Parachute Infantry Regiment, fire during a squad live-fire exercise at Fort A.P. Hill, Virginia, March 14, 2018.
(U.S. Army photo by Spc. John Lytle)
Or, take a battle above ground. The Marines think they may be denied conventional radio communications in a war with China or Russia. Any robots helping them will only be able to communicate within a short range or by using lasers. Lasers, obviously, become short range communications when there are a lot of obstructions, like dense foliage or hills, in the way.
So, these robots will also need to complete moment-by-moment tasks while also coordinating their actions whenever they can communicate. All of this requires that the robots keep a constantly updating list of what tasks need completed, what humans haven’t been checked on in a while, and what areas are safe or unsafe for the robots to operate in.
MIT’s PR2 robot loads beers into the cooler of a “turtle” waiter bot as part of a program to improve robots’ ability to coordinate their actions in challenging environments.
Or, as MIT graduate student Ariel Anders said, “These limitations mean that the robots don’t know what the other robots are doing or what the other orders are. It forced us to work on more complex planning algorithms that allow the robots to engage in higher-level reasoning about their location, status, and behavior.”
From an MIT article about the team’s paper:
“These uncertainties were reflected in the team’s delivery task: among other things, the supply robot could serve only one waiter robot at a time, and the robots were unable to communicate with one another unless they were in close proximity. Communication difficulties such as this are a particular risk in disaster-relief or battlefield scenarios.”
So, yeah, at MIT, a beerbot is never just about beer. And the actual tech underlying these social-media-friendly beerbots is actually necessary for the less sexy but more vital missions, like disaster relief. And, potentially, it could even save the lives of troops under fire or wounded service members in the next few years or decades.
Let the military have its beerbots. And, if they sometimes use them for beer instead of medical supplies, well, they would’ve found a way to get drunk anyways.