How Routing Strategies Impact Urban Emissions

TL;DR

This work addresses how GPS routing apps influence urban CO2 emissions and introduces TraffiCO2, a simulation framework that couples real mobility data, routing APIs (OSM and TomTom), a microscopic traffic simulator, and the HBEFA3 emission model to estimate edge-level emissions. The approach systematically varies the fraction of vehicles following routing suggestions and introduces perturbations to non-R routes, revealing that both extreme adoption (all or none) elevate emissions, while about 40–70% following with some randomness minimizes total emissions and distributes them more evenly. Key findings show emissions are highly spatially heterogeneous and that TomTom routing tends to underperform OpenStreetMap in total emissions but still benefits from partial routing; randomization further reduces both emissions and travel time. The framework offers a novel, open tool for evaluating routing strategies and informs next-generation routing principles that balance individual needs with urban well-being across different cities and pollutants.

Abstract

Navigation apps use routing algorithms to suggest the best path to reach a user's desired destination. Although undoubtedly useful, navigation apps' impact on the urban environment (e.g., carbon dioxide emissions and population exposure to pollution) is still largely unclear. In this work, we design a simulation framework to assess the impact of routing algorithms on carbon dioxide emissions within an urban environment. Using APIs from TomTom and OpenStreetMap, we find that settings in which either all vehicles or none of them follow a navigation app's suggestion lead to the worst impact in terms of CO2 emissions. In contrast, when just a portion (around half) of vehicles follow these suggestions, and some degree of randomness is added to the remaining vehicles' paths, we observe a reduction in the overall CO2 emissions over the road network. Our work is a first step towards designing next-generation routing principles that may increase urban well-being while satisfying individual needs.

Figures (10)

• Figure 1: Schema of the TraffiCO$_2$ simulation framework. (a) The city is split into squared tiles using scikit-mobility scikitmob. (b.1) Real data are used to estimate mobility flows (OD matrix) within the city. (b.2-b.3) A trip is created by selecting at random an origin-destination pair from the OD matrix and two edges on the road network. (b.5) Some routing algorithm is used to convert each trip into a path on the road network. (c) Steps b.1-b.5 are repeated $N$ times ($N$ = number of vehicles) to obtain a multiset of routed paths. (d) A traffic simulator (SUMO) is used to simulate the urban traffic generated by the routed paths (e).
• Figure 2: Examples of routed paths between an origin and destination pair according to OSM (blue), TT (red), DR with $w=5$ (orange), and DR with $w=1$ (black).
• Figure 3: The Complementary Cumulative Distribution Functions (CCDFs) of the CO$_2$ (in mg) emitted on the roads (averaged across 10 repetitions) by the vehicles of the generated urban traffic $S(\overline{D}_{i}^{(R)}, \mathcal{T})$, for OSM (a) and TT (b). Colours represent routed paths with increasing percentage of $R$-routed vehicles. The inset plot zooms on the distributions' tail.
• Figure 4: Gini index of the CO$_2$ distribution (a) and total CO$_2$ emissions (b) varying the percentage of $R$-routed vehicles, for OSM (blue) and TT (red). In the error bars, points indicate the average Gini index (a) and the total CO$_2$ (b) over ten simulations with different choices of $R$-routed vehicles chosen uniformly at random. Vertical bars indicate the standard deviation.
• Figure 5: The difference in the total CO$_2$ emitted on each road (in mg per meter of road) when: (a) none of the vehicles is OSM-routed and 50% of them are ($\mathcal{E}_0^{\text{\tiny (OSM)}}(e) - \mathcal{E}_5^{\text{\tiny (OSM)}}(e)$, $\forall e \in E$); (b) all vehicles are OSM-routed and 50% of them are ($\mathcal{E}_{10}^{\text{\tiny (OSM)}}(e) - \mathcal{E}_5^{\text{\tiny (OSM)}}(e)$, $\forall e \in E$). Red roads indicate a positive difference; blue ones indicate a negative difference.