Deliberative Self-Organizing Traffic Lights with Elementary Cellular Automata

Jorge L. Zapotecatl, David A. Rosenblueth, and Carlos Gershenson
Complexity
Volume 2017 (2017)
May 30, 2017

Abstract

Self-organizing  traffic lights have shown considerable improvements compared to traditional  methods in computer simulations. Self-organizing methods, however, use  sophisticated sensors, increasing their cost and limiting their deployment.  We propose a novel approach using simple sensors to achieve self-organizing  traffic light coordination. The proposed approach involves placing a computer  and a presence sensor at the beginning of each block; each such sensor  detects a single vehicle. Each computer builds a virtual environment  simulating vehicle movement to predict arrivals and departures at the  downstream intersection. At each intersection, a computer receives  information across a data network from the computers of the neighboring  blocks and runs a self-organizing method to control traffic lights. Our  simulations showed a superior performance for our approach compared with a  traditional method (a green wave) and a similar performance (close to  optimal) compared with a self-organizing method using sophisticated sensors  but at a lower cost. Moreover, the developed sensing approach exhibited  greater robustness against sensor failures.

Related publications