Lane-Level Road Map Construction considering Vehicle Lane-Changing Behavior

In recent years, the construction of lane-level road maps has received extensive attention from industry and academia. It been widely studied because this kind map provides foundation for much research, such as high-precision navigation, driving behavior analysis, traffic analysis. Trajectory-based crowd-mapping is an emerging approach to construction. However, major problem that existing methods neglect modeling trajectory distribution in longitudinal direction road, which significantly impacts precision. Thus, article proposes a two-stage method based on vehicle lane-changing model road’s lateral distributions simultaneously. first stage, behaviors are extracted trajectories. second lane extraction established using weighted constrained Gaussian mixture hidden Markov estimate parameters (e.g., counts centerline) each cross section. Then accurate continuous centerlines can be constructed accordingly. The proposed verified data collected crowdsourced platform named Mapillary. results show construct information satisfactorily.