Monitoring physical distancing for crowd management: Real-time trajectory and group analysis.

Pouw, Caspar A S; Toschi, Federico; van Schadewijk, Frank; Corbetta, Alessandro · 2020 · DOAJ

DOI: 10.1371/journal.pone.0240963

archive: archived pipeline: cataloged verified

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

Summary

This study addresses the challenge of monitoring physical distancing compliance in crowded public spaces, specifically train stations, during the COVID-19 pandemic. The authors aim to develop a privacy-respectful, real-time analysis framework capable of distinguishing between pedestrians who violate distancing rules and those who are part of family groups, which are exempt from such constraints. The research is motivated by the need for efficient crowd management tools that can process large volumes of trajectory data without compromising individual privacy or incurring prohibitive computational costs. The researchers utilized pedestrian tracking data collected from 19 commercial overhead sensors at Platform 3 of Utrecht Central Station in the Netherlands. These sensors provided high-resolution, anonymous trajectory data with sub-meter accuracy. The study compares pre-pandemic data (May 2019) with data collected during the pandemic (May 2020). The core methodological innovation is a scalable graph-based framework where pedestrians are represented as nodes and their pairwise interactions as weighted edges. The edge weights record the frequency of distance events within specific bins, effectively creating a discrete counterpart to the Radial Distribution Function (RDF). This "additive" approach allows for the incremental identification of family groups—defined as individuals maintaining consistent short distances over time—and the detection of "Corona events," where non-family members breach the 1.5-meter Dutch distancing threshold. The analysis reveals significant changes in crowd behavior and density. Pre-pandemic, crowd densities frequently exceeded 1 pedestrian per square meter, whereas during the pandemic, maximum densities dropped to approximately 0.3 pedestrians per square meter. The RDF analysis showed a marked depletion in short-distance interactions during the pandemic compared to pre-pandemic levels. By applying the graph-based method, the authors successfully disentangled family-group interactions from rule violations. In a sample of 75 passengers during the pandemic, only three individuals were identified as distance offenders, whereas in a comparable pre-pandemic sample, about one-third of people stood closer than 1.5 meters to others. The framework effectively automated the identification of groups and violations in real-time, validating its utility for monitoring adherence to social distancing measures. The significance of this work lies in providing a computationally efficient, privacy-compliant tool for real-time crowd management. The proposed graph-based framework not only aids in enforcing physical distancing but also offers a generalizable method for analyzing group dynamics and motion patterns in public transport facilities. Furthermore, the statistical observables derived from the study, such as exposure times and distance distributions, can inform the calibration of contact tracing applications by providing empirical data on the relationship between proximity, duration, and contagion risk in urban settings.

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.

• perception reaction time