Modeling pedestrian-vehicle interactions at unsignalized mid-block crosswalk under automated vehicle mixed traffic flow conditions: An agent-based approach

The introduction of automated vehicles (AVs) is expected to improve the safety and traffic efficiency of the transportation system. However, the impacts of AVs on pedestrian-vehicle interactions (PVIs) remain unclear, especially under the mixed traffic flow condition with AVs and conventional vehicles (CVs). The objective of this paper is to investigate PVIs under AV mixed traffic flow at unsignalized mid-block crosswalk (UMC) where the right of way competition exists and the traffic conditions are more complex. An agent-based simulation model is proposed to evaluate the safety and traffic efficiency under a diverse range of different scenarios. Some agents’ behavioral parameters are calibrated from trajectory and video data collected at two sites in Nanjing. The proposed model is also validated by comparing post-encroachment time (PET) and pedestrian waiting time outputted from simulation with those extracted from empirical data. The simulation results suggest that the introduction of AVs has limited effect on both safety and efficiency levels at UMC under the low AV penetration rate. Compared to the negotiated yielding strategy of AVs, the conservative yielding strategy performs better in terms of pedestrian safety, but worse in improving traffic efficiency. Moreover, the results reveal that pedestrian and vehicle volumes, the proportion of pedestrians familiar with AVs, and the proportion of risk-prone pedestrians could significantly affect the safety and efficiency levels at UMC. Lastly, we provide some practical recommendations for determining treatments for mid-block crosswalks with AV mixed traffic flow to improve pedestrian safety and traffic efficiency.