Distributed intelligence in robots

In new research published today in Nature Communications, a group of European researchers have come out with a new kind of modular robot.

Many robots can complete a task by themselves, but don't work well as a team. On the other hand, scientists are now building machines that use distributed intelligence like that of a bee hive to work together, but these bots don't work well as independent components.

The researchers at Université Libre de Bruxelles wanted to build a versatile robot that could act independently but, when it encounters other like robots, surrender itself to a leader-bot, which they've called the "brain" unit. It's a sort of air traffic controller for the other robots that can arrange them as needed.

They create an artificial nervous system where each bot acts as a different neuron. But if the brain unit is damaged, the system has been designed to compensate. Three previous "neuron" units are deputized to act like one part of the brain unit, detaching from the dead node before reforming to carry on the work. The researchers hope to forge a world that looks past purpose-built bots and towards one where a single robot platform could work for every task ... as long as you can tell the individual components how to fall in place right.

"Our vision is that, in the future, robots will no longer be designed and built for a particular task," they write. "Instead, we will design composable robotic units that give robots the flexibility to autonomously adapt their capabilities, shape and size to changing task requirements."