Search

During winter, road conditions play a crucial role in traffic flow efficiency and road safety. Icy, snowy, slushy, and wet road conditions reduce tire friction and affect vehicle stability which could lead to dangerous crashes. To keep traffic operations safe, cities spend a significant budget on winter maintenance operations such as snow plowing and spreading of salt and sand. This paper proposes a methodology for automated winter road surface condition classification using convolutional neural network (CNN) and the combination of thermal and visible light cameras. As part of this research, 4244 pairs of visible light and thermal images are captured from pavement surfaces and classified into snowy, icy, wet, and slushy surface conditions. Two single-stream CNN models (visible light and thermal streams) and one dual-stream CNN model are developed. The average F1-Score of dual-stream model is 0.866,0.935,0.985, and 0.888 on snowy, icy, wet, and slushy, respectively. The weighted average F1-Score is 0.94. Key words: winter road surface condition monitoring, winter road maintenance, convolutional neural network, sensor fusion, thermal camera. En hiver, les conditions routieres ont un impact crucial sur l'efficacite de la circulation et la securite routiere. Les conditions glacees, enneigees, boueuses et mouillees sur la route reduisent le frottement des pneus et ont une incidence sur la stabilite du vehicule, ce qui pourrait entrainer des accidents dangereux. Pour assurer la securite de la circulation, les villes consacrent un budget important aux operations d'entretien hivernal, comme le deneigement et l'epandage de sel et de sable. Cet article propose une methodologie de classification automatisee des conditions de la surface de roulement d'hiver a l'aide d'un reseau neuronal convolutif (RNC) et de la combinaison de cameras thermiques et de cameras a lumiere visible. Dans le cadre de cette recherche, 4244 paires d'images de lumiere visible et d'images thermiques sont capturees a partir des surfaces de chaussee et classees en conditions de surface neigeuse, glacee, humide et boueuse. Deux modeles RNC a flux unique (lumiere visible et flux thermiques) et un modele RNC a double flux sont developpes. Le score F1 moyen du modele a double flux est de 0,866, 0,935,0,985 et 0,888 sur neige, glace, mouillee et neige fondante, respectivement. Le score F1 moyen pondere est de 0,94. [Traduit par la Redaction] Mots-cles : surveillance de l'etat de la surface des routes l'hiver, entretien des routes l'hiver, reseau neuronal convolutif, fusion de capteurs, camera thermique., 1. Introduction In winter, road surface condition greatly affects traffic efficiency and safety, especially in countries that experience long winters and harsh climates. The literature discusses the relationship between road [...]

A number of traffic safety studies have investigated temporal variations in road safety indicators such as crash frequency, confirming that such variations exist. This paper examined whether speeding is more likely on weekend nights relative to all other times of the day by directly comparing speeding during weekends and weekdays. To this end, we analyzed a sample of local streets, in Montreal, for which speed data were collected automatically using traffic analyzer sensors. We found that, interestingly, weekend speeding was less likely to occur during night hours, whereas it was more likely to occur during evening and midday hours. Among other findings, the results indicated that one-way streets and those having a speed limit of 50 km/h were slightly less prevalent in speeding on weekends. Our results can be useful in designing road safety interventions, including publicity campaigns and police enforcement, which aim at reducing speeding behaviours. Key words: speeding behaviour, weekends, weekdays, speed limits, temporal variation. Un certain nombre d'etudes sur la securite routiere ont examine les variations temporelles des indicateurs de securite routiere, comme la frequence des accidents, ce qui confirme l'existence de telles variations. Cet article a examine si les exces de vitesse sont plus probables la nuit de fin de semaine par rapport a toutes les autres heures de la journee en comparant directement les exces de vitesse pendant les fins de semaine et les jours de semaine. A cette fin, nous avons analyse un echantillon de rues locales, a Montreal, pour lesquelles des donnees de vitesse ont ete recueillies automatiquement a l'aide de capteurs pour l'analyse de la circulation. Fait interessant, nous avons constate que les exces de vitesse en fin de semaine etaient moins susceptibles de se produire pendant les heures de nuit, alors qu'ils etaient plus susceptibles de se produire le soir et le midi. Entre autres conclusions, les resultats ont indique que dans les rues a sens unique et celles qui ont une limite de vitesse de 50 km/h les exces de vitesse les fins de semaine etaient un peu moins frequents. Nos resultats peuvent servir a la conception des interventions en matiere de securite routiere, y compris des campagnes de publicite et l'application de la loi, visant a reduire les exces de vitesse. [Traduit par la Redaction] Mots-cles: exces de vitesse, fins de semaine, jours de semaine, limites de vitesse, variation temporelle., Introduction Active travel modes safety is a major concern in many urban settings. For example, more than 15% of Canadian road fatalities were pedestrians for the period 2013 to 2017, [...]

This paper analyses bicycle ridership patterns using a unique database of automated bicycle counts from approximately 40 locations in five North American cities and along the Route Verte in Quebec. The cities involved in this study are Montreal, Ottawa, Portland, San Francisco, and Vancouver. Count data show that the bicycle volume patterns at each location can be classified as utilitarian, mixed utilitarian, recreational and mixed recreational. Study locations classified into each of these categories are found to have consistent hourly and weekly traffic patterns, despite important differences between these cities in terms of factors such as weather, size, and urban form. Expansion factors for each location type are presented by hour and day of the week. There were differences in seasonal patterns of bicycle activity between the study locations, so different monthly expansion factors are presented for each city. Finally, some traffic volume characteristics are presented for comparison purposes.