Authors: Philippe Apparicio*, Urbanisation Culture Société, Jérémy Gelb, INRS Urbanisation Culture Société, Fabrice Dumont, INRS Urbanisation Culture Société
Topics: Transportation Geography, Urban Geography, Spatial Analysis & Modeling
Keywords: Real-time traffic, Air Pollution, Road traffic noise, cycling
Session Type: Paper
Start / End Time: 4:40 PM / 6:20 PM
Room: Galerie 3, Marriott, 2nd Floor
Presentation File: No File Uploaded
In studies on cyclists' exposure to air pollution or noise during trips through cities, it is difficult to obtain and integrate data about real-time traffic. These data are therefore often replaced by proxies such as the road typology. Moreover, few free tools are currently available to count vehicles in a video, in particular when the camera is in motion. However, this parameter is essential to constructing non-biased exposure models. This study proposes a new methodology and some tools designed to tackle this problem. Trips collected by bicycle in Montreal, Mexico City and Paris during one of our studies are used to test this approach.
Videos of trips are collected by action cameras (Garmin Virb) and constitute the raw material. A web platform developed during the project is used by operators to count vehicles in videos and to classify them according to a pre-established typology. The platform makes it possible to test the concordance between the results obtained by many operators in the same video for different time resolutions.
In the study, two operators counted the vehicles in videos of the three cities with a high level of concordance (higher than 85%). Differences were significant only for the first few attempts when counting rules were not precisely defined for operators.
Finally, these traffic measures were incorporated into our exposure models. They clearly enhanced the capacity of the models to predict exposure to air pollution and noise and modified the level of significance of the other variables.