Compared with human-like robots that are being built to comprehend emotion, perform complex tasks, and outsmart people on quiz shows, a trio of bug-like machines designed by Harvard University engineers with a handful of crude sensory abilities hardly seems to merit the word “intelligence.”
But this seemingly ragtag team of robots is one example of a powerful alternate approach to tackling complex problems: instead of trying to build one smart machine, build a bunch of average ones. Inspired by termites, which can accomplish astonishing feats of engineering despite having no foreman telling them what to do, the researchers decided to see if droves of simple robots could carry out a complex task by working together.
In a study published in the journal Science on Thursday, the team from the Wyss Institute for Biologically Inspired Engineering at Harvard described what can best be called a kind of collective intelligence. A team of robots given a picture of a structure they’re supposed to build can do so not by following a preset plan, but by sensing what exists in the environment and following a few rules.
They do not need to communicate with one another, they do not need a central computer to tell them what steps to follow and when, and they don’t need to know how the structure has changed since they laid down the last brick.
“Every robot is doing its own thing and doesn’t know anything about the rest,” said Justin Werfel, a scientist at the Wyss. “If they break or they’re lost, the rest of them can keep on working without changing anything about what they’re doing.”
The world of swarm robotics is challenging the popular idea that the future of robotics is exclusively about smart machines. In fact, asked what she might have done differently if she were to start the project over again, Harvard graduate student Kirstin Petersen said she might make the robots even simpler.
“If I built up a new system, I might make the robots less capable,” Petersen said. “But make more of them.”