— A "swarm" of simple-minded robots that teams up to move an object too heavy for them to manage individually has been demonstrated by robotics researchers.
The robots cannot communicate and must act only on what they can see around them. They follow simple rules to fulfil their task - mimicking the way insects work together in a swarm.
The robots were developed by Marco Dorigo at the Free University of Brussels, Belgium, along with colleagues at the Institute of Cognitive Science and Technology in Italy and the Autonomous Systems Laboratory and Dalle Molle Institute for the Study of Artificial Intelligence, both in Switzerland.
"In the future we might have robots that actively seek help from others when they come up a problem they can't solve alone," says Dorigo, "For example if a robot can't climb an obstacle without tipping over it might go back and get others to climb over as a group."
In experiments, six of the cylindrical robots were able to drag an object across the floor of a room. Working autonomously, they locate and assemble around the object and either grab hold of it directly or of another robot nearby, before slowly dragging it towards a target.
A video shows the six Swarm-bot robots gradually transporting a object lit with red LEDs over to a large white target.
Another video clip, shown at 10 times normal speed, shows a larger team of robots working together to map out a path from a red object and a blue target. This strategy is necessary because none of the bots can see far enough to work out the route between the object and its target for themselves.
Each Swarm-bot is 19 centimetres high, has a rotating turret, a claw-like gripper and moves using a combination of caterpillar tracks and wheels. Each also has a basic computer and is loaded with the same software.
The simple rules laid out in this software allow the robots to perform complex actions as a group. A swarm of ants uses a similar strategy to tackle difficult jobs like carrying a large object.
The rules preloaded onto the Swarm-bots were "evolved" to suit the particular task and incorporated genetics-based algorithms and a detailed 3D simulation (see Nuclear reactors 'evolve' inside supercomputers).
"In the object transport scenario they search for a red object and grasp onto it," explains Dorigo. "When they do that they also change colour from blue to red." This means a cluster of bots is "connected" to the object. When the bots cannot see any more blue - meaning they are all linked together - they start dragging the object towards its target.
The robots can adjust their caterpillar tracks, to ensure they are all pulling in the right direction. "Each robot has a traction sensor inside that detects all the external forces on it," explains Dorigo. A robot uses its sensor to identify any conflicting forces, and then changes direction accordingly.
Dorigo is now working on a swarm of robots that could operate in a human environment. "It is called Swarmanoid and will have three different kinds of robots," he explains. Some robots will be able to crawl along like Swarm-bots, others will be able to climb walls, and others still will be able to fly, he says.