In a world of Darwinian selfishness, it is comforting to find that there is still a place for teamwork. Ants, bees and even bacteria are amongst the many organisms that display sophisticated kinds of cooperative behaviour as a survival strategy. Putting many 'hands' to a single task such as foraging for food can be much more efficient than letting each individual fend for itself. Now scientists and engineers who seek efficient solutions to complex tasks are learning valuable lessons from insects.
Ants are a classic example of social insects, which work together for the good of the colony. A colony of ants finds new food sources by sending out foragers who explore the surroundings more or less at random. If it finds food, a forager will return to the colony, laying a pheromone trail as it goes -- a trail that other ants can follow back to the food.
But this is not in itself a perfect stratagem. The successful forager does not know the most direct way back to the colony, and so there is a risk that it may send its colleagues on a circuitous route to the food. Another forager might subsequently find a better route -- but how would others know to take it in preference? Indeed, colonies can sometimes get stuck on an unnecessarily long path.
Fortunately, shorter trails are more regularly refreshed with new pheromone, and so are more likely to stay marked than long trails. In effect, the ants have the potential to select the best route.
This kind of scheme can help computer scientists find the best solution to many a thorny and time-consuming problem, Eric Bonabeau of the Santa Fe Institute in New Mexico, USA, and colleagues report in
Such problems can be solved by simulating the journeys of a swarm of trail-laying ants, Bonabeau's team suggests. Each 'virtual ant' marks the journey with a 'pheromone' that characterizes how short its tour was. Good routes become more attractive to other ants than poor routes. Simulating pheromone evaporation at a steady rate ensures that the colony does not get trapped onto mediocre solutions that prevent them from looking for better ones.
The researchers call this an 'Ant Colony Optimization (ACO) algorithm', an example of the approach to computational problem-solving known as Swarm Intelligence. ACO is currently being explored for planning petrol truck routes in Switzerland.
Ants generally move over open terrain. Engineers are finding it useful to place their virtual foraging insects inside networks. This can help to identify, for example, the best way of routing signals through a communications network. The foragers can adapt to cope with complications such as congestion: journeys that take a long time as a result of such delays are not strongly reinforced, so better, alternative routes are quickly found. The constant searching and feedback of the swarm makes this approach, 'Ant Colony Routing', highly flexible and responsive to changes such as removal or addition of channels to the network. British Telecom is exploring routing algorithms of this kind.
One of the most intriguing applications of Swarm Intelligence is to the design of real robotic systems in which the robots are tiny devices that act collectively. A swarm of micro-robots could adapt its behaviour more quickly than a single one, and could operate using simpler programming principles.