Modeling bidirectional mixed flow of pedestrians, bicycles and E-mopeds at signalized crosswalks
针对信号交叉口行人、自行车和电动助力车混合双向流，开发了基于智能体的模型，利用南京CBD七处交叉口的轨迹数据验证，发现隔离式人行横道和“靠右”规则均能提升通行速度、降低冲突风险。
Despite extensive research on pedestrian-vehicle interactions in shared spaces, the behavioral dynamics and movement patterns among heterogeneous vulnerable road users (VRUs) in mixed flows at signalized crosswalks remain underexplored. To address this gap, this study develops an agent-based model (ABM) for bidirectional mixed flow of pedestrians, bicycles and electric mopeds (e-mopeds) at signalized crosswalks. The model incorporates heterogeneous personal space and interactive behaviors to capture the influence of individual characteristics and environmental factors on movement dynamics. Using trajectory data from 2,133 pedestrians, 890 bicycles, and 3,337 e-mopeds collected at seven urban crosswalks in Nanjing CBD, China, we analyze speed distribution, spatial relation and route choice across different user types. Results demonstrate that the proposed model can accurately reproduce dynamic interactions and collision avoidance behaviors under varying traffic conditions. Both segregated crosswalk and crosswalk with ‘keep right’ rule improve crossing speeds and reduce conflict risks under high traffic volume. The segregated crosswalk outperforms crosswalk with ‘keep right’ rule on pedestrian speed enhancement and conflict mitigation, while crosswalk with ‘keep right’ rule show superior performance for bicycle and e-moped speed optimization and conflict reduction.