Inspired by the collective intelligence of termites, researchers have created an autonomous robotic construction crew -- no supervisor, no communication, no advanced planning needed -- an important proof of concept for scalable, distributed artificial intelligence.
"Normally, at the beginning, you have a blueprint and a detailed plan of how to execute it, and the foreman goes out and directs his crew, supervising them as they do it," Harvard's Justin Werfel says in a press release. "In insect colonies, it's not as if the queen is giving them all individual instructions. Each termite doesn't know what the others are doing or what the current overall state of the mound is."
In Namibia, for example, millions of termites working for over a year have built an 8-foot-tall mound of soil -- a "lung" for their nest underground. Without a coordinated strategy, they rely on a kind of implicit communication called "stigmergy" -- observing each others' changes to the environment and acting accordingly.
So Werfel and colleagues leveraged stigmergy to design algorithms that reflect termite behavior, and then implemented these algorithms in a set of construction robots. With their TERMES system, an assemblage of simple robots can build complex, 3D structures -- towers, castles, and pyramids -- without any central command, prescribed roles, or detailed plans. They just need the ability to sense a brick or bot nearby to make their next move.
Each robot in the construction team is equipped with just four sensors and three actuators. Here are the things TERMES robots can do:
Although each one follows just simple rules, together they exhibit intelligent behavior, doing "something really complicated as a group, without a supervisor," Harvard's Radhika Nagpal says, "without everybody discussing explicitly what's going on." If one robot breaks, or has to leave, it doesn't affect the others. Despite the loss (or not knowing about it), they keep working by following the cues of those robots that remain. This means the same instructions can be executed by five robots or 500. According to a Harvard press release:
It is currently more common for robotic systems to depend on a central controller. These systems typically rely on an "eye in the sky" that can see the whole process or on all of the robots being able to talk to each other frequently... but as the numbers of robots and the size of their territory increase, these systems become harder to operate. In dangerous or remote environments, a central controller presents a single failure point that could bring down the whole system.
In the future, these sorts of termite-inspired robots -- which are expendable and especially useful in places we might find "dirty, dangerous and dull," as Werfel says -- could lay sandbags before a flood and build shelters after an earthquake, and even perform simple construction tasks underwater or on Mars. "If you could send a team of robots to go build the habitat as the first step -- that's the really long-term vision," Werfel adds.
Here's a nifty video of the termite robots at work:
The work was presented at 2014 American Association for the Advancement of Science (AAAS) meeting in Chicago and published in Science last week.
Read more: Harvard press release, AAAS release, Science, New Scientist
Images: Eliza Grinnell, Harvard School of Engineering and Applied Sciences
This post was originally published on Smartplanet.com