Brainy ants
30 November 2000 13:56
By Tornado Staff
That ants are industrious is no novelty, but the news that the tiny animals organise their work with intelligence may come as a surprise. The method they employ is ingenious to the point that Marco Dorigo, an Italian researcher at Iridia, Artificial intelligence institute of the Université Libre de Bruxelles, decided to devise an algorithm to reproduce their mechanisms. With his ant algorithms he aims to improve the efficiency of human endeavour.
“Ants’ organisation structures” explains Dorigo “are something quite admirable, as those of termites, bees and wasps. Artificial intelligence researchers have been interested in this sort of behaviour for the past 30 years, and today interest is as keen as ever”. To grasp the ants’ secrets it is necessary to get biologists and mathematicians on the case. The former carry out observation and data collection; the latter attempt to translate the results of research into a logical process.
“From lab trials” carries on Dorigo “it appears that, when ants have to chose between two paths, they never fail to identify the shortest one. Their method is simple: ants which pioneer new paths leave traces of a pheromone, a smell which informs the ants that follow whether the path is encumbered by obstacles or not. On the basis of this principle, new applications can be developed, which similarly leave traces behind, so to optimise a system’s performance”.
The algorithms can be applied in the most varied fields: from automation systems to the management of production processes. The most important uses, though, are within Internet and logistics. It is possible to feed in intelligent agents which surf relentlessly the network of networks and identify congested nodes. Artificial ants are also able to modify routing, i.e. that mechanism which directs information packets from one node to the next. This is known as Ant Colony Optimization (Aco).
“An email which has to go from Milan to Brussels doesn’t necessarily get there faster by following the shortest path” says Dorigo “if in Frankfurt traffic is at a standstill, then it might be quicker to go via San Francisco. Considering that data on the web moves at the speed of light, priority has to be given to finding a free path rather than running fewer kilometres”. The agents though are not yet operative because it is first necessary to seek the approval of large Internet providers. Those wishing to use them on an Intranet are free to do so: Dorigo’s algorithms are public domain. This is what Netherlands-based Unilever did, and uses them to optimise operation sequencing for certain machinery.
Related stories
|
| | Data
25,600
Tech investments
From our Online Data Service
|
16,571
VC-backed companies
From our Radar
|
Recent Deals
For information on Europe's most extensive database on technology funding click here!
|  |
Events
Press Releases
|
