Navigation services amplify concentration of traffic and emissions in our cities

TL;DR

This study investigates how GPS navigation services shape urban traffic and CO2 emissions by embedding public-service route recommendations within a realistic, data-driven urban traffic simulator. Using a SUMO-based digital twin built from OpenStreetMap networks and GPS-derived origin-destination flows across Florence, Milan, and Rome, the authors vary navigation adoption rate $r$ and compare routed versus non-routed drivers. They find a universal pattern of route conformity: higher adoption concentrates traffic and emissions on fewer roads, increasing local inequality, while low adoption can reduce CO2, but benefits fade beyond city- and service-dependent thresholds. The marginal link between reduced route diversity and emissions follows an exponential form $ ext{Δ}E_r= ext{α}e^{-eta ext{ΔC}_r}+ ext{γ}$, with city-specific $eta$ values and a stabilization of this relationship above traffic thresholds (e.g., $N=10{,}000$ for Florence, $20{,}000$ for Milan, $35{,}000$ for Rome). The framework is open-source and adaptable for policy testing, enabling evaluation of interventions to mitigate emergent externalities from algorithmic routing and human–AI coevolution in cities.

Abstract

The proliferation of human-AI ecosystems involving human interaction with algorithms, such as assistants and recommenders, raises concerns about large-scale social behaviour. Despite evidence of such phenomena across several contexts, the collective impact of GPS navigation services remains unclear: while beneficial to the user, they can also cause chaos if too many vehicles are driven through the same few roads. Our study employs a simulation framework to assess navigation services' influence on road network usage and CO2 emissions. The results demonstrate a universal pattern of amplified conformity: increasing adoption rates of navigation services cause a reduction of route diversity of mobile travellers and increased concentration of traffic and emissions on fewer roads, thus exacerbating an unequal distribution of negative externalities on selected neighbourhoods. Although navigation services recommendations can help reduce CO2 emissions when their adoption rate is low, these benefits diminish or even disappear when the adoption rate is high and exceeds a certain city- and service-dependent threshold. We summarize these discoveries in a non-linear function that connects the marginal increase of conformity with the marginal reduction in CO2 emissions. Our simulation approach addresses the challenges posed by the complexity of transportation systems and the lack of data and algorithmic transparency.

Figures (15)

• Figure 1: Routes recommended by navigation services.(a) Example of an origin-destination pair where suggestions by a navigation service (TTF) overlap considerably (63.44% on average). (b) Example where suggestions differ considerably (average overlap of 17.17%). Different navigation services generally suggest different routes for the same origin-destination pair. This variation occurs because the services rely on different criteria and possess diverse historical data on traffic conditions.
• Figure 2: Impact of navigation services on route diversity and CO2 emissions.(a-c) Service adoption rate ($r$) versus route diversity at high traffic loads in Florence, Milan, and Rome. The dashed line represents the no-impact scenario ($r{=}0\%$). Markers indicate the average route diversity over ten simulations with different random choices of $s$-routed vehicles. Squares refer to the navigation service ecoTT, which employs eco-routing. Vertical bars indicate the standard deviation. The inset plots zoom on the range $r = 0\%, \dots, 50\%$, where route diversity slightly increases. Increased adoption of navigation services reduces route diversity, with only minor fluctuations among navigation services. (d-f) Service adoption rate ($r$) versus average CO2 emissions per vehicle at high (filled markers) and low (empty markers) traffic loads. The dashed line represents the no-impact scenario ($r=0\%$). Markers indicate the average CO2 emissions over ten simulations with different random choices of $s$-routed vehicles. Squares refer to ecoTT. Vertical bars indicate the standard deviation. At high traffic loads, when $r$ is low, CO2 emissions decrease considerably; when $r$ exceeds a certain city- and service-dependent threshold, the benefits plateau and, in some cases, even reverse.
• Figure 3: Road usage of navigation services.(a, b) Route distribution of trips in Milan with TTF adoption rate $r{=}0\%$ (a) and $r{=}100\%$ (b). Darker edge colours indicate higher traffic concentration. When $r{=}0\%$, traffic is more spatially uniform than traffic when $r{=}100\%$, at which there is a concentration of routes on fewer edges. (c) Difference in road usage between $r{=}100\%$ and $r{=}0\%$. Blue edges indicate where the total adoption of TTF reduces traffic compared to the scenario with no impact; red edges indicate where it concentrates traffic more.
• Figure 4: Relationship between route diversity and CO2 emissions.(a-c) The relationship between the marginal change in route diversity ($\Delta D_r$) and the marginal change in CO2 emissions ($\Delta E_r$) for TTF in Florence, Milan, and Rome. We find similar results for the other navigation services. The points are shaded from light grey to black, representing the service adoption rate (from $r{=}10\%$ to $r{=}100\%$). (d-f) The relationship between $\Delta D_r$ and $\Delta E_r$ for all navigation services. Regions where $\Delta D_r$ or $\Delta E_r$ are negative are highlighted in grey. The black dashed line represents the exponential decay fit for each scenario. At low adoption rates, slight increases in route diversity lead to substantial reductions in CO2 emissions. As $r$ increases, small reductions in route diversity result in moderate CO2 reductions. However, as $\Delta D_r$ further increases, $\Delta E_r$ decreases, indicating a diminishing return effect. This pattern is consistent across all cities and navigation services.
• Figure S1: Service adoption rate versus route diversity at low (a-c) and high (d-f) traffic loads in Florence, Milan, and Rome. The dashed line represents the no-impact scenario (adoption rate $r=0\%$). In the error bars, points indicate the average route diversity over ten simulations with different choices of $s$-routed vehicles chosen uniformly at random. Vertical bars indicate the standard deviation. The inset plots zoom on the range $r = 0\%, \dots, 50\%$, where route diversity slightly increases. The consistency of this universal pattern is evident across different cities and varying traffic conditions.