Capacity-Aware Deep Learning for Generalizable Traffic Volume Estimation Across Links and Cities
2026-07-27 • Machine Learning
Machine LearningArtificial Intelligence
AI summaryⓘ
The authors developed a method to estimate how much traffic is on roads using commonly available data like speed measurements, road details, and weather, instead of relying heavily on fixed traffic sensors. They trained a model using limited sensor data and tested it both on new roads in the same area and on entire new cities. Their approach treats this as a challenge of predicting traffic patterns in unfamiliar places, and they improved accuracy by factoring in road capacity and typical traffic usage. Experiments showed their method works better than previous ones when dealing with new locations and sparse data.
traffic volume estimationlink-level learningprobe speed profilesspatial generalizationtopological descriptorstraffic-theoretic constraintsspatial distribution shiftsupervised learningroad capacityregime-aware utilization
Authors
Léo Hein, Giovanni De Nunzio, Aurélie Pirayre, Laurent Najman
Abstract
Network-wide traffic volume estimation typically relies on propagating measurements from fixed sensors, making performance highly dependent on sensor density and limiting deployment in sparsely instrumented networks. We propose a link-level learning framework that estimates hourly traffic volumes from widely available territorial data only, including probe speed profiles, road and topological descriptors, along with weather observations. A supervised local mapping is learned from sparse sensor measurements and evaluated under two generalization settings: intra-network (unseen links within the training network) and inter-network (unseen city). This formulation frames traffic volume estimation as a spatial out-of-distribution generalization problem under sparse supervision. To enhance spatial robustness, we introduce a capacity-aware formulation that models volume as the product of a link-specific structural capacity and an hourly regime-aware utilization ratio, embedding traffic-theoretic constraints directly into the learning process. Extensive experiments in both generalization settings demonstrate that the proposed structural constraints consistently outperform a state-of-the-art baseline under spatial distribution shift.