We all know how to spell. There's nothing very special about that. And teaching a robot how to spell isn't all that difficult, either. But Edward Macdonald, a masters student at Georgia Tech, has built a small army of 15 Khepera robots that can spell out words without even communicating with each other. And they do it by expending the least amount of energy.

Part of his final thesis project, Macdonald's robots know only their own position and the relative position of each of the other robots thanks to a camera suspended above the platform they rest on and a Wi-Fi network. "With this basic data, the robots use a mathematical algorithm to calculate how to make the shape they've been asked to make, with the least amount of traveling," according to Techland's Giles Turnbull.

Watch the demonstration below, in which the robots spell out GRITS for Georgia Robotics and Intelligent Systems. When the operator moves one robot out of the group, the rest rearrange themselves to make up for its loss. The result has been described, accurately, as a "delightful techno ballet."

Watch other Videos on the Technology Channel.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.