How Will the Presence of Autonomous Vehicles Affect the Equilibrium State of Traffic Networks?

Negar Mehr, Roberto Horowitz

Abstract

Connected and autonomous vehicles (AVs) are capable of maintaining shorter headways and distances when they form platoons, potentially increasing traffic network capacities and boosting mobility. This paper investigates the validity of this impact under selfish routing behavior in networks with mixed autonomy (both regular and autonomous vehicles). Using a nonatomic routing game with inelastic demands and modified U.S. Bureau of Public Roads (BPR) delay models, we show that the resulting Wardrop equilibrium is not necessarily unique in networks with mixed autonomy. While we identify conditions under which total network delay is guaranteed not to increase, we also demonstrate that counter-intuitive behaviors can occur: total delay may grow as network autonomy increases. Specifically, for single O/D pair networks with homogeneous road asymmetry, the total delay is unique and non-increasing. However, for heterogeneous asymmetry, uniqueness is lost, and performance can degrade. We further bound these performance degradations in homogeneous networks.

Introduction

The deployment of autonomous vehicles (AVs) is widely expected to revolutionize urban mobility. By communicating and reacting faster than human drivers, AVs can maintain shorter headways, effectively "platooning" to increase the throughput of existing infrastructure. However, most traffic networks operate under the principle of selfish routing: individual drivers choose paths to minimize their own travel time, leading to a Wardrop Equilibrium where no driver can benefit from a unilateral route change.

This research addresses a critical question: if we replace a fraction of human-driven vehicles with selfishly routed AVs, is the network guaranteed to improve? Or does the increased capacity on specific links tempt drivers into routes that ultimately congest the system for everyone, similar to the famous Braess’s Paradox?

Modeling Mixed Autonomy

The study employs a nonatomic routing game framework. Unlike standard models where road capacity is a fixed constant, this model treats capacity as a function of the "autonomy ratio" on each specific link. A key parameter is the capacity asymmetry ($\mu$), defined as the ratio between the capacity of a road used only by regular vehicles versus one used only by AVs.

The Wheatstone bridge network topology used to demonstrate how adding capacity or autonomy can lead to counter-intuitive delay increases.

The authors distinguish between two types of networks:

  1. Homogeneous Networks: Where AVs provide the same relative capacity boost across every road in the network.
  2. Heterogeneous Networks: Where different roads benefit to varying degrees from autonomy technology.

The Paradox of Autonomy

A central finding is that increasing AV penetration is mathematically analogous to adding capacity to specific links. In a heterogeneous network, this can trigger a version of Braess's Paradox. For example, in a Wheatstone bridge topology (Figure 2), increasing the capacity of the central "bridge" link through AV platooning can lead all drivers to shift their routes in a way that increases the total travel time for every participant.

Furthermore, the paper reveals that the standard assumption of a unique equilibrium delay no longer holds in mixed-autonomy settings. Depending on the initial distribution of traffic, the system may settle into different equilibrium states with varying total delays for the same overall fraction of AVs.

Key Results and Performance Bounds

The research demonstrates that in heterogeneous networks, a system with 100% AV penetration can actually suffer higher total delays than a system with 0% AVs. This performance degradation is captured by a new metric, the "Price of Vehicle Autonomy" ($\eta$), which measures the worst-case delay increase caused by the introduction of AVs.

Total network delay (J) as a function of the autonomy fraction (α), illustrating non-uniqueness and potential performance degradation.

As shown in Figure 3, for a given autonomy fraction ($\alpha$), there can be a range of possible total delays ($J_{min}$ to $J_{max}$), highlighting the non-uniqueness of the equilibrium. While the total delay is guaranteed to be a non-increasing function of autonomy in single O/D networks with homogeneous asymmetry, this stability vanishes in more complex or heterogeneous scenarios.

For homogeneous networks, the authors established an upper bound on this degradation. For standard polynomial delay functions of degree 4, the total delay in a mixed-autonomy network is at most ~2.15 times the delay of the non-autonomous case.

Conclusion

The study suggests that the mere presence of AVs is not a panacea for traffic congestion. Because selfish routing can exploit capacity gains in ways that harm the collective, the deployment of AVs must be managed carefully. To realize the full mobility benefits of autonomy, cities may need to move beyond simple infrastructure upgrades toward active traffic management strategies, such as dynamic tolling or centralized routing, to prevent the paradoxes inherent in selfish mixed-autonomy traffic.

Summary updated

Abstract

It is known that connected and autonomous vehicles are capable of maintaining shorter headways and distances when they form platoons of vehicles. Thus, such technologies can result in increases in the capacities of traffic networks. Consequently, it is envisioned that their deployment will boost the network mobility. In this paper, we verify the validity of this impact under selfish routing behavior of drivers in traffic networks with mixed autonomy, i.e. traffic networks with both regular and autonomous vehicles. We consider a nonatomic routing game on a network with inelastic (fixed) demands for the set of network O/D pairs, and study how replacing a fraction of regular vehicles by autonomous vehicles will affect the mobility of the network. Using the well known US bureau of public roads (BPR) traffic delay models, we show that the resulting Wardrop equilibrium is not necessarily unique even in its weak sense for networks with mixed autonomy. We state the conditions under which the total network delay is guaranteed not to increase as a result of autonomy increase. However, we show that when these conditions do not hold, counter intuitive behaviors may occur: the total delay can grow by increasing the network autonomy. In particular, we prove that for networks with a single O/D pair, if the road degrees of asymmetry are homogeneous, the total delay is 1) unique, and 2) a nonincreasing continuous function of network autonomy fraction. We show that for heterogeneous degrees of asymmetry, the total delay is not unique, and it can further grow with autonomy increase. We demonstrate that similar behaviors may be observed in networks with multiple O/D pairs. We further bound such performance degradations due to the introduction of autonomy in homogeneous networks.
BibTeX
@article{mehr2020how,
  author = {Negar Mehr and Roberto Horowitz},
  title = {{How Will the Presence of Autonomous Vehicles Affect the Equilibrium State of Traffic Networks?}},
  journal = {IEEE Trans. Control. Netw. Syst.},
  year = {2020},
  url = {https://openreview.net/forum?id=368BbQNYXOe},
}
Figure