Recognition of Intersection Traffic Regulations from Crowdsourced Data

In this paper, a new method is proposed to detect traffic regulations at intersections using GPS traces. The knowledge of rules for regulated locations can help various location-based applications in the context Smart Cities, such as accurate estimation travel time and fuel consumption from starting point destination. Traffic map features, however, are surprisingly still largely absent maps, although they do affect flow which, turn, affects vehicle idling intersections, consumption, CO2 emissions, arrival time. addition, mapping them surveying equipment costly any update process has severe constraints. This fact precisely motivation study. Therefore, its objective propose an automatic, fast, scalable, inexpensive way identify type intersection control (e.g., lights, stop signs). A based on summarizing collective behavior crossing proposed. modification well-known clustering algorithm used stopping deceleration episodes. These episodes then categorize into four possible categories (p1: free flow, p2: without events, p3: only one event, p4: more than event). percentages crossings each class per arm, together with other speed/stop/deceleration extracted trajectories, features classify arms according their (dynamic model). classification results dynamic model compared those static model, where OpenStreetMap. Finally, hybrid also tested, combination used, which outperforms two models. For three models, variants feature vector tested: associated single arm (one-arm model) another neighboring same (all-arm methodology was tested datasets show that all-arm models perform better single-arm accuracy 95% 97%.