lights organize traffic
Technology Research News
is notoriously difficult to predict because it's a complicated system involving
many vehicles, many people's behavior, and constant changes. If a single
car takes more time than expected to start up at a green light, the cars
behind will be delayed. Put a few such delays together and you have traffic
jam, never mind traffic-altering variables like weather, roadworks, sports
events, concerts and holidays.
Traffic lights that adapt to traffic flow are currently in use,
but often consist of relatively expensive equipment including a control
center that receives information from traffic lights and sends back appropriate
A researcher from the Free University of Brussels (VUB) in Belgium
has devised a way to allow traffic lights to self-organize to improve traffic
flow. The method, which taps the self-organizing principles of social insects,
does away with central control.
Traditionally, traffic lights are optimized using computational
methods that find appropriate periods and phases of traffic lights, said
Free University of Brussels researcher Carlos Gershenson. But traffic flow
is less an optimization problem than an adaptation problem because traffic
flows and densities change constantly, he said.
In Gershenson's relatively simple simulations, traffic lights were
able to self-organize and adapt to changing traffic conditions, reducing
wait times and increasing average speeds compared to a pair of traditional
traffic-management methods. An increase in average speed shows that cars
are spending less time stopped at lights.
The method is based on a fairly simple approach used in England
for coordinating traffic at isolated intersections, said Gershenson. A red
light keeps track of the number of cars approaching it and multiplies that
number by time increments. When the result reaches a certain threshold,
the traffic light changes. The more cars approaching or waiting at a red
light, the faster it will turn to green.
The traffic light pattern encourages cars to form platoons, which
improve average speeds. This method works well when traffic is light, but
in heavy traffic the method does not work because the threshold is reached
very quickly, and the lights change constantly.
Gershenson adapted the method to make it work in heavy traffic by
adding a constraint: a green light cannot be switched red before a minimum
Other researchers have developed traffic management schemes that
call for traffic lights to communicate with each other.
Under the Free University of Brussels scheme, the traffic lights
cannot communicate directly with each other, but reacting to local traffic
density allows them to coordinate indirectly. "They communicate indirectly
through their environment, as many social insects do," said Gershenson.
"The environment of the traffic lights [is] the cars."
The method allows the traffic lights to exploit the environmental
information -- cars -- to self organize. "There is no need of the central
command center, communication between agents, nor hierarchies," said Gershenson.
It remains to be seen which approach works better in which situations,
Gershenson's simulations represented traffic flow in a 10-by-10
grid of one-lane roads of alternating directions. They allowed more cars
flowing in a particular direction to simulate rush-hour traffic. Cars could
go straight or turn at an intersection.
The simulation varied the number of initial cars among 10, 20, 40,
80, 160, 360, 640, 1,280 and 2,560. Gershenson took the averages of 20 runs
lasting 10,000 time increments each.
Gershenson's next steps are to put the idea through more complicated
simulations that consider pedestrians and left hand turns, run a pilot study,
and compare it with current adaptive traffic management systems.
The system could be applied practically in less than two years,
according to Gershenson. The research was funded by the University.
Timeline: > 2 years
TRN Categories: Multiagent Systems; Applied Technology
Story Type: News
Related Elements: Technical paper, "Self-Organizing Traffic
Lights," posted on the arXiv physics archive at xxx.lanl.gov/abs/nlin.AO/0411066;
self-organizing traffic lights simulation posted at homepages.vub.ac.be/~cgershen/sos/SOTL/SOTL.html
26/February 2, 2005
Nano bridge builds logic
Braille display drives
Adaptive lights organize
the physics of the very small
photocells harvest infrared
Oil and water drive
makes cheap sensors
Metals speed transparent
Plastic memory retains
Research News Roundup
Research Watch blog
View from the High Ground Q&A
How It Works
News | Blog
Buy an ad link