MIT is Building Autonomous Collaborating Robot Herds

These robot “fleets” can work together to accomplish menial tasks.

Recommendations are independently chosen by Reviewed’s editors. Purchases you make through our links may earn us a commission.

I had a mild existential crisis when I sat down to write this post, and it had nothing to do with the usual response to a news story about artificially intelligent robots (insert stale quip about The Terminator or The Matrix). It had to do with snark.

Each morning, I sit down here at my desk in Cambridge, Massachusetts, and begin looking for story ideas. When I find one—let’s say, a story about a beer fridge that can only be opened by Canadians—I scoop out the most essential news elements, make sure I understand all the details, and then brainstorm some humorous angles (i.e., “Oh Canada, with your… with your… maple leaves… and your… gravy fries”). You know: snark.

You see, snark is the low-hanging fruit of web journalism. It’s the sardonic dialect of needy millennials competing for the attention of an attention-deficient generation. And I am as guilty as the next writer in using it.

But then this happened—this story—and my usual response hit a brick wall. I felt like Gob (for all you Arrested Development fans), spiraling into a helpless, panicky fit of insecure blather: Should, should, should, should I, should I, should I just, should I just?

For once, snark just wouldn’t do—not for something so immensely compelling, and so positively frightening. Here’s why (mercifully without snark):

This week, researchers at MIT unveiled a system that allows “herds” of artificially intelligent robots to autonomously collaborate with one another. According to a report, the system “combines simple control programs to enable fleets of robots… to collaborate in unprecedented ways.”

"The system enables fleets of robots to collaborate in unprecedented ways.”

*(Furiously resisting snark)

The program uses several iRobot Creates—programmable robots with the same chassis as the Roomba vacuum cleaner—to maneuver and relocate a series of small objects. Most importantly, these robot fleets—or “multiagent systems”—collaborate with each other to both find these objects and physically move them.

If you’ve ever seen a Roomba, you know there’s very little threatening about them, but when they begin to work in tandem with one another, and when you extrapolate the exponential potential of robotic A.I., you have one frightening little vacuum. (Ugh! Snark—couldn’t resist).

“In [multiagent] systems, in general, in the real world, it’s very hard for them to communicate effectively,” Christopher Amato, of MIT’s Computer Science and Artificial Intelligence Lab, told MIT News. “If you have a camera, it’s impossible for the camera to be constantly streaming all of its information to all the other cameras. Similarly, robots are on networks that are imperfect, so it takes some amount of time to get messages to other robots, and maybe they can’t communicate in certain situations around obstacles.”

Given the complex demands of AI, such a feat is truly remarkable.

To get around this problem, researchers programmed the robots with three inputs: first, a series of algorithms to govern behavior; second, data for executing those programs in particular environments; and third, a mechanism for assessing different outcomes—a balance between negative and successful performance.

What that all means is, the robots are able to receive instructions for completing a task, interact with their physical environment, and collaborate with other robots to complete said task.

Given the complex demands of AI, such a feat is truly remarkable, and the researchers at MIT deserve all the credit for developing such a system. But you can’t help but wonder: What happens when roboticists combine this level of collaboration, with this level of communication, and this level of agility?

You get a Terminator. A Terminator is what you get.

Snark is dead. Long live snark.

Up next