Sign In

Communications of the ACM

ACM News

Group Thinker: Researcher Gets $2.9 Million to Further Develop Swarm Intelligence

View as: Print Mobile App Share:
The European Research Council is giving researchers in Belgium $2.9 million to further study the potential that swarm intelligence holds for improving information technology and robotics.


Swarm intelligence is a branch of artificial intelligence that attempts to get computers and robots to mimic the highly efficient behavior of colony insects such as ants and bees. Ants, for example, use pheromone trails to mark the routes they use to find food. The more traversed trails develop an accumulation of pheromone that attracts new ants, whereas pheromones deposited on paths less traveled will evaporate over time.

With an eye on the potential that swarm intelligence holds for the development of information technology and robotics, the European Research Council awarded a $2.9-million ERC Advanced Grant Tuesday to Marco Dorigo to help the research director for the Belgian Funds for Scientific Research and co-director of IRIDIA (the Free University of Brussels's artificial intelligence lab) further his work engineering swarm intelligence systems.

Phone systems use a similar approach to routing calls, using bits of information as "virtual pheromones" that reinforce paths through less congested areas of a network, researchers Eric Bonabeau and Guy Théraulaz noted in Scientific American's February 2008 special issue on robots. Dorigo and his colleagues have applied this philosophy to the Internet and managed to outperform all other data-traffic routing methods, the authors added.

Swarm intelligence systems promise to provide an alternate way of designing systems that have greater autonomy and self-sufficiency, "relying on direct or indirect interactions among simple individual agents," according to Bonabeau and Théraulaz. The difficulty is that researchers still lack a detailed understanding of the inner workings of insect swarms. Computer scientists have been unable to identify the specific rules by which individuals in a swarm interact, making it difficult for them to develop more advanced software that mimics this behavior, the authors wrote.

From Scientific American
View Full Article


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account