Analyzing Cycling Behavior and Traffic State Interactions at Unsignalized Intersections using Bird's-Eye View Video Data

This study investigates cycling behavior in relation to the traffic state of the forward area using multi-modal trajectories from bird's-eye view video data. Cycling is recognized as a sustainable mode of transportation, with cities promoting urban designs to encourage it. However, the rise in cyclist numbers has led to more accidents, especially in mixed traffic areas such as unsignalized intersections. Despite its importance, cycling behavior remains under-researched compared to car traffic due to the complexity of modeling non-lane-based movements and data collection limitations. Existing models focus on specific interactions, but robust and comprehensive models that consider various traffic situations are still lacking. This study addresses this gap by analyzing cycling behavior based on the surrounding traffic state, particularly forward speed and lateral fluctuations. Occupancy inside the intersection is considered a traffic state factor that may affect behavior. Using bird's-eye view video data from Munich, multi-modal trajectories are extracted and labeled by traffic mode based on the YOLO algorithm. Correlation analysis reveals significant differences between lateral and forward speed fluctuations, indicating two types of cyclists, with the majority detouring for collision avoidance rather than waiting. Forward speed and occupancy also show different significant correlations for the cyclists tending to detour in congested versus less congested situations. Insights from this study can help develop robust models for estimating the impact of cycling behavior on other road users, adaptable to different traffic scenarios and infrastructure designs, and from other users to bicyclists. Applications for this research include microscopic models and vehicle collision avoidance models.     «

This study investigates cycling behavior in relation to the traffic state of the forward area using multi-modal trajectories from bird's-eye view video data. Cycling is recognized as a sustainable mode of transportation, with cities promoting urban designs to encourage it. However, the rise in cyclist numbers has led to more accidents, especially in mixed traffic areas such as unsignalized intersections. Despite its importance, cycling behavior remains under-researched compared to car traffic du...     »