The march of the kilobots

“A THOUSAND-ROBOT SWARM” sounds like the title of a 1950s science-fiction B movie. It is actually, though, the title (or, strictly, part of the title) of a paper in this week’s Science. Indeed, the paper in question, by Michael Rubenstein of Harvard University and his colleagues, describes what might, in the hands of a film-maker, be an even scarier idea: a swarm whose members can co-ordinate their actions.

To be precise, Dr Rubenstein’s ’bot swarm (above) has 1,024 members (2¹⁰ being a conveniently binary number), known apparently without irony as kilobots. Each is a rigid-legged tripod that moves around by vibrating. Kilobots communicate with infra-red light, which can reflect off the table Dr Rubenstein uses for his experiments, and are programmed with three types of behaviour.

One is edge-following, which allows a ’bot move along the edge of a cluster. The second is gradient-formation, which lets it know how many other ’bots a signal has been relayed through, and thus gives it information about the location of these ’bots and the shape of the cluster it is in. The third is localisation, which means it can agree a system of co-ordinates with its neighbours, so that they can measure distances between themselves.

To form a shape, each ’bot is told what the desired outcome is, and four—which act as “seeds”—are also informed of their starting-points. After that, it is just a question of waiting for the machines to hustle their way around until all agree the goal has been reached.

At the moment, the simple shapes (such as letters of the alphabet, stars and even a spanner) which Dr Rubenstein gets his ’bots to make are of little practical value. The emergence of order from chaos, though, is impressive, and more sophisticated swarms whose members had, say, arms as well as legs, might well be able to do useful things.