Teams of mindless robots can achieve complex tasks without communicating with each other, say robotics experts in Canada, who were inspired by ants' behaviour. Armies of such cheap, expendable robots might one day help build a base on Mars-or simply mow your lawn.
Non-communicative behaviour in ants is observed when they combine their efforts to carry large pieces of food, such as leaves, says Ronald Kube, a robotics expert at the Edmonton Research Park in Alberta. He says he has now mimicked the same sort of distributed intelligence in a gang of worker robots.
If you watch a particular ant, its behaviour seems chaotic and sometimes counterproductive, says Kube. But the "team" as a whole displays a form of intelligence despite the lack of central control.
This is attractive to people who build robots because such distributed systems tend to be very robust. It is a lot cheaper and easier to build a large number of simple robots than build one expensive complex robot to do the same job. A centrally controlled system can be brought down by a single failure. "But who cares if I lose two or three hundred of these, as long as the job gets done?" asks Kube.
His challenge was to find a job in which some robots could be lost without preventing the task being completed. "We had to pick a problem that was hard enough that it couldn't be carried out by a single robot," says Kube, who worked on the project with Eric Bonabeau, then at the Santa Fe Institute in New Mexico.
The team's six robots are intrinsically stupid, says Kube. Their mission is to search an arena for food, which in this case is a light source on a box. The box is too heavy for a single robot to push. Once they reach the food source, each robot has to establish if it is facing a second light on the ceiling at the other end of the arena, which represents the nest. If they are, they try to push the food to the nest. If not, then they wander around in a counterclockwise fashion until they find the food again and then re-evaluate whether or not to push.
Although the behaviour of individuals appears chaotic (see the videos at www. cs.ualberta.ca/~kube/crip.cgi), the robots accomplish their task using only the position of the box as the means of indirect communication. Eliminating the need for robots to communicate gets rid of the thorny and hardware-hungry problem of orchestrating their messaging.
Chris Melhuish, a robot expert at the University of West England in Bristol, says finding the simple rules that produce this kind of behaviour is tricky-but by mimicking nature, at least we know these rules exist.
Author: Duncan Graham-Rowe
New Scientist issue 29th January 2000
PLEASE MENTION NEW SCIENTIST AS THE SOURCE OF THIS STORY AND, IF PUBLISHING ONLINE, PLEASE CARRY A HYPERLINK TO : http://www.newscientist.com
Source: Robotics and Autonomous Systems (vol 30, p 85)