CMStatistics 2022: Start Registration
View Submission - CMStatistics
B1268
Title: Spatial statistical calibration for linear network data: The analysis of traffic volumes Authors:  Andrea Gilardi - University of Milano - Bicocca (Italy) [presenting]
Riccardo Borgoni - University of Milano-Bicocca (Italy)
Jorge Mateu - University Jaume I (Spain)
Abstract: Analyzing traffic volumes at the street network level represents a crucial step to improving transport planning protocols and developing effective road safety interventions. A common practice to estimate traffic figures involve manual counts with ad-hoc cameras or automatic counts with road-fixed sensors (e.g., inductive loops and spirals). Unfortunately, these traditional techniques have several limitations due to their limited spatial coverage and high economic costs of installation and maintenance. For these reasons, in the last years, several authors developed statistical methods to derive counts from traffic information collected using geo-referenced mobile sensors (e.g., smartphones and sat-navs). Mobile sensors have several advantages over traditional instruments, but they underestimate real flows. For these reasons, we developed a spatial statistical calibration technique to combine accurate counts from fixed cameras and extensive GPS mobile data to estimate traffic flows, re-adapting the methodology to the linear network context. We also propose an algorithm that can be used to optimise the size and the spatial allocation of fixed sensors in an urban environment by evaluating their importance for spatial calibration. The suggested methodology is exemplified by considering data collected in the City of Leeds (UK).