|
Abstract:
|
While modern traffic signals are increasingly sensor based, many older traffic signals are controlled by predetermined timings. These timings are rarely perfect and often cause unnecessary delays in traffic and increased congestion in cities. Converting existing traffic signals to use sensors will increase the efficiency of the traffic signal, but this option is often too expensive to consider. Updating signals with better timings could be an inexpensive way to increase traffic signal efficiency. There is no good method for choosing signal timings and evolutionary algorithms are a promising alternative to setting them by hand. Because of the dynamic nature of vehicle traffic, swarm intelligence might provide a good alternative to the probabilistic traffic models that are often used in traffic simulations.
|
