Using Real Sensors Data to Calibrate a Traffic Model for the City of Modena

Bachechi, Chiara; Rollo, Federica; Desimoni, Federico; Po, Laura · 2020 · Crossref

DOI: 10.1007/978-3-030-39512-4_73

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of creating accurate, dynamic traffic simulations for urban environments, specifically within the city of Modena, Italy. Motivated by the TRAFAIR project’s goal to monitor and forecast urban air quality, the authors seek to overcome the limitations of static traffic models, which only provide average conditions during peak hours. Instead, they aim to develop a micro-simulation model that captures the dynamic evolution of traffic flows throughout the day and year. The study utilizes real-time data from over 300 induction loop sensors located at traffic-light-controlled junctions in Modena, which had previously been used only for local traffic light control rather than broader analysis. The methodology employs the open-source software SUMO (Simulation for Urban Mobility) and OpenStreetMap (OSM) to build a cost-effective micro-simulation model. In this model, individual vehicles are simulated based on routes generated from sensor data. "Calibrators" are placed near each physical sensor to inject traffic flow matching real-time vehicle counts and average speeds recorded every minute. To validate the model, the authors compare simulated data from virtual detectors against real sensor measurements using Dynamic Time Warping (DTW) distance and average difference metrics. Tests were conducted on seven days in November 2018. The evaluation revealed that more than 20% of calibrators were "not aligned" with real data. The authors identified three primary causes for these discrepancies: inaccuracies in the OSM road network regarding lane directions, the accumulation of "fake jams" due to long simulation durations, and erroneous input from sensors reporting zero vehicles throughout the day. To resolve these issues, the authors tested different model configurations. They first attempted to exclude unreliable calibrators and sensors with zero counts, which improved some results but left 58 calibrators misaligned. The most effective solution involved splitting the 24-hour simulation into eight sub-simulations of three hours each. This approach prevented the accumulation of unnecessary vehicles that caused artificial congestion. Consequently, the number of not-aligned calibrators dropped dramatically to 2.0% (5 out of 241). The remaining misalignments were associated with the same junctions that had consistently failed in previous tests, suggesting persistent structural or data issues rather than simulation artifacts. The study concludes that splitting simulations into shorter intervals and filtering out non-functional sensor data significantly enhances the realism and accuracy of urban traffic models. This calibrated model can reliably reproduce vehicle counts at sensor locations and infer flows in areas with nearby sensors. The authors suggest that future improvements should include Origin-Destination matrices to better model routes in areas lacking sensor coverage. This work demonstrates that leveraging existing infrastructure sensor data with refined simulation techniques can provide robust tools for traffic management and environmental monitoring.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• traffic density

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Methodological Resource: validation psychometrics