Achieving Level 5 autonomous driving in typical urban environments requires a comprehensive understanding of the interactions among diverse road users. These include not only human-driven vehicles, but also parked or stopped vehicles, motorcycles, and bicycles, all of which contribute to the complexity of urban traffic. This study employs a multi-agent simulation framework to examine how the coexistence of such heterogeneous traffic participants affects overall traffic flow. Each agent is modeled using a combination of a risk potential model and a kinematic model, allowing for the analysis of mutual influences arising from changes in traffic participants and environmental conditions. The findings provide insights into the challenges of fully autonomous driving in mixed urban traffic and highlight the importance of accounting for diverse road user behaviors in traffic flow modeling.
KITAZAWA et al. (Wed,) studied this question.