Analysis of Road Safety Speed from Floating Car Data

Intelligent Transportation Systems aims at improving efficiency and safety of the transportation system by acting either on vehicle performances or assisting the driver with information on vehicle and traffic status. Although digital road graphs are available to derive quantitative parameters that describe the road geometry, the information provided usually includes speed limits and repetition of road signs. On the other hand, a huge amount of data is available on individual vehicle speeds and trajectories collected as Floating Car Data (FCD) but they are not combined with road parameters to derive information on how drivers perceive the infrastructure and behave when traveling on it. In the paper, a methodology is presented to evaluate the consistency between drivers' behavior and a theoretical safety speed determined from road geometry. The azimuth profile is progressively built for a road layout, based on the geometry described by a digital graph. Consecutive elements with the constant azimuth variation are identified as circular curves and their radii are computed by circle fitting. The safety speed with respect to longitudinal stability is estimated. The obtained safety speed is then compared to the distribution of speeds observed from about 200 million FCD collected on the regional road network of Latium. The obtained results permit to individuate critical points of the network in terms of road safety.