Realizing Wardrop Equilbria with Real-Time Traffic Information

TL;DR

This paper investigates how real-time traffic information can be used to achieve Wardrop equilibria, analyzing different driver behavior scenarios and control systems through simulations on a two-route example.

Contribution

It demonstrates that real-time data and anticipation can help approximate Wardrop equilibria in traffic networks, addressing phase transitions and oscillations.

Findings

Selfish routing causes oscillations in travel times.

Control systems can stabilize traffic flow near equilibrium.

Anticipation-based advice improves route choice stability.

Abstract

A Wardrop equilibrium for multiple routes requires equal travel time on each path used. With real-time traffic data regarding travel times, it is important to analyze how to use the information provided. In particular, can a Wardrop equilibrium be realized? Simulations using the three-phase model on a two-route example are presented to answer this question. One route (the main line) is a two-lane highway with a stalled vehicle in the right lane and the other route is a low-speed bypass. For a critical incoming flow, a phase transition between free flow and congested flow near the stalled vehicle is observed, making this a challenging example. In the first scenario, drivers choose routes selfishly on the basis of current travel times. The result is strong oscillations in travel time because of the inherent delay in the information provided. The second scenario involves a hypothetical control system that limits the number of vehicles on the main line to prevent the free-flow to congested-flow phase transition by diverting sufficient flow to the bypass. The resulting steady state is neither a Wardrop equilibrium nor a system optimum, but an intermediate state in which the main-line travel time is less than on the bypass but the average for all vehicles is close to a minimum. In a third scenario, anticipation is used as a driver-advice system to provide a fair indicator of which route to take. Prediction is based on real-time data comparing the number of vehicles on the main line at the time a vehicle leaves the origin to the actual travel time when it reaches the destination. Steady states that approximate Wardrop equilbria, or close to them, are obtained.