Generating a Lane-Specific Transportation Network Based on Floating-Car Data

Future applications in ITS and automated driving require high precise digital maps including a lane-specific transportation network. The paper presents a method for estimating lane center lines based on vehicle trajectories from floating-car data. Kernel density estimation was applied for estimating lane center lines. The floating-car dataset is based on measurements on three different road types (urban 3-lane freeway, urban arterial, rural 2-lane freeway) using different low-cost GNSS receivers (GPS data logger and several smartphone GPS positioning apps). As reference, some test runs were conducted with high precise D-GPS measurement equipment. The longitudinal and lateral positioning errors were analyzed within a roadway and trip based distance analysis. The final results show deviations less than 0.14 m in median between measured and estimated lane center lines. This accurate estimation of lane center lines allows a generation of lane-specific transportation networks based on common floating-car data.