Tuesday, June 29, 2010

Taking Traffic Control Lessons — From Ants

If humans took their cues from ants, they might spend less time in traffic.
When opposing streams of leafcutter ants share a narrow path, they instinctively alternate flows in the most efficient way possible. Studying how ants manage this could provide the basis for a system of driverless cars running on ant traffic algorithms.
"They never get stuck in traffic," said Audrey Dussutour, a University of Sydney entomologist. "We should use their rules. I’ve been working with ants for eight years, and have never seen a traffic jam — and I’ve tried."
People have long been fascinated with the ability of ants to organize colonial activities in patterns as sophisticated as any urban engineer’s megalopolis blueprint. In recent years, scientists have turned ant traffic flows into algorithms applicable to data transmission and vehicular traffic.
Dussutour, whose earlier work showed that leafcutter ants organize themselves into separate and tightly-regulated streams of load-carrying and unburdened individuals when traveling in opposite directions on wide paths, was curious about their dynamics on narrow paths such as the tip of a treebranch — the ant equivalent of a one-lane road.
In the latest findings, published in the February issue of the Journal of Experimental Biology, Dussutour’s team found that ants leaving the colony automatically gave right-of-way to those returning with food. Of the returning ants, some were empty-mandibled — but rather than passing their leaf-carrying, slow-moving brethren, they gathered in clusters and moved behind them.
This seemingly counterintuitive strategy — when stuck behind a slow-moving truck, are you content to slow down? — actually saved them time.
"Leafcutters paths in particular look very much like car traffic," said Dussutour. "There’s a lot of times on the highway when you’re stuck behind a truck, and sometimes overtaking it is not optimal."
The results are an example of how individual behaviors optimized to serve a collective good can ultimately benefit the individual as well. If humans would let a network take the wheel, these principles might manage our own congested thoroughfares.
"We essentially would have to hand over control of the vehicle to a collectively intelligent system that would move all vehicles from their source to destination," said Marcus Randall, a Bond University software mathematician. People would be reluctant, he said, but "accidents would be virtually non-existent and travel would become much more efficient.
If ants in the experiment behaved like the average human driver, they’d routinely run head-first into each other, causing insect versions of pile-ups and gridlock. Dussutour’s team calculated that patience reduced the average delay experienced by an individual ant crossing a crowded three-meter bridge from 64 to 32 seconds.
"One dominating factor in human traffic is egoism," said University of Zoln traffic flow theorist Andreas Schadschneider.
"Drivers optimize their own travel time, without taking much care about others. This leads to phantom traffic jams which occur without any obvious reason. Ants, on the other hand, are not egoistic."
Another way of understanding the difference between human and ant navigation decisions, he said, comes from optimization theory. In human traffic, "the user optimum is relevant, whereas in ant traffic it is the system optimum, which can be quite different," and produces a different set of behaviors.
Guiding the individual ants’ decisions is their inherited, colony-serving programming and on-the-ground traffic updates, acquired from an immersive cloud of information that takes the form of pheromone trails and physical contact. Though scientists have studied and pheromones for decades, the latter exchange is less understood.
"We have good evidence that encounters between inbound and outbound workers are important," said study co-author Sam Beshers, an entomologist at the University of Illinois at Urbana-Champaign. "It’s not just that they’re managing the traffic flow. They’re managing the information flow, too.
That’s about all we know, but it’s potentially very important."
An experimental navigational system called Inter-Vehicle Communication tries to emulate this, with on-board navigation computers exchanging data as they pass each other and roadside base stations. It’s yet to be deployed in real-world conditions, though, reflecting the difficulty in replacing a culture and infrastructure of solitary driving.
A compromise, said Schadschneider, may be systems that improve communication between drivers and cars. "This has already been achieved by new devices which transmit information about abrupt velocity decreases to the following cars, which then start to brake automatically, before the driver even realizes the need to brake," he said.
Beshers is optimistic about the potential of driverless cars running on ant traffic algorithms, but cautious about the timeline of their acceptance. Embracing such a system, he said, "assumes that humans could agree on an upper speed limit, which has never yet happened."
Citation: "Priority rules govern the organization of traffic on foraging trails under crowding conditions in the leaf-cutting ant Atta colombica." By Dussutour, A., Beshers, S., Deneubourg, J. L. and Fourcassié, V.. Journal of Experimental Biology, Vol. 212 Issue 4, Feb. 15, 2009.
Source :  http://www.wired.com/wiredscience/2009-/02/anttraffic/#ixzz0sEBMAST7

No comments:

Post a Comment