Traffic flow on realistic road networks with adaptive traffic lights

TL;DR

This paper models traffic flow on urban networks using cellular automata, demonstrating that adaptive traffic lights informed by upstream and downstream data improve average travel times and reduce fluctuations.

Contribution

It introduces a cellular automata-based model for urban traffic with adaptive signals informed by real-time data, comparing different control strategies on real and hypothetical networks.

Findings

Adaptive traffic lights improve average network performance.

Joint upstream-downstream control reduces travel time fluctuations.

Model validated on Melbourne's road network.

Abstract

We present a model of traffic flow on generic urban road networks based on cellular automata. We apply this model to an existing road network in the Australian city of Melbourne, using empirical data as input. For comparison, we also apply this model to a square-grid network using hypothetical input data. On both networks we compare the effects of non-adaptive vs adaptive traffic lights, in which instantaneous traffic state information feeds back into the traffic signal schedule. We observe that not only do adaptive traffic lights result in better averages of network observables, they also lead to significantly smaller fluctuations in these observables. We furthermore compare two different systems of adaptive traffic signals, one which is informed by the traffic state on both upstream and downstream links, and one which is informed by upstream links only. We find that, in general, both the mean and the fluctuation of the travel time are smallest when using the joint upstream-downstream control strategy.