Failure of classical traffic flow theories: Stochastic highway capacity and automatic driving

TL;DR

This paper reviews the failure of classical traffic theories to explain empirical traffic breakdowns and explores how automatic driving vehicles influence highway capacity and traffic stability, revealing potential deterioration even with small vehicle percentages.

Contribution

It establishes a connection between empirical stochastic highway capacity and the impact of automatic driving vehicles using probabilistic simulations within three-phase traffic theory.

Findings

Automatic driving vehicles can both decrease and increase traffic breakdown probability.

Even 5% automatic vehicles can deteriorate traffic flow performance.

Automatic vehicles may increase breakdown risk despite satisfying string stability conditions.

Abstract

In a mini-review [Physica A {\bf 392} (2013) 5261--5282] it has been shown that classical traffic flow theories and models failed to explain empirical traffic breakdown -- a phase transition from metastable free flow to synchronized flow at highway bottlenecks. The main objective of this mini-review is to study the consequence of this failure of classical traffic-flow theories for an analysis of empirical stochastic highway capacity as well as for the effect of automatic driving vehicles and cooperative driving on traffic flow. To reach this goal, we show a deep connection between the understanding of empirical stochastic highway capacity and a reliable analysis of automatic driving vehicles in traffic flow. With the use of simulations in the framework of three-phase traffic theory, a probabilistic analysis of the effect of automatic driving vehicles on a mixture traffic flow consisting of a random distribution of automatic driving and manual driving vehicles has been made. We have found that the parameters of automatic driving vehicles can either decrease or increase the probability of traffic breakdown. The increase in the probability of traffic breakdown, i.e., the deterioration of the performance of the traffic system can occur already at a small percentage (about 5$\%$) of automatic driving vehicles. The increase in the probability of traffic breakdown through automatic driving vehicles can be realized, even if any platoon of automatic driving vehicles satisfies condition for string stability.