Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces a link-level learning framework for estimating hourly traffic volumes using only widely available territorial data, circumventing the limitations of sparse sensor networks. By framing the problem as a spatial out-of-distribution generalization challenge, the authors enhance estimation accuracy through a capacity-aware formulation that incorporates traffic-theoretic constraints. Experimental results show that this approach significantly outperforms state-of-the-art methods, particularly in scenarios involving unseen links and cities.
Traffic volume estimation can be achieved with remarkable accuracy using only territorial data, even in sparsely monitored networks.
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.