Impact of merging behavior and control algorithm on mixed traffic flow of human driven vehicles and autonomous vehicles

With the development of autonomous driving technology, autonomous vehicles (AVs) have become increasingly prevalent in real-world. While merging behavior and control algorithms are known to affect traffic, their combined effect especially under timid and yielding strategies remains unexplored. This paper studies the impact of merging behavior of human driven vehicles (HVs) and control algorithm of AVs on the mixed traffic flow of HVs and AVs. We consider four cases, in which the merging behavior and control algorithm are either timid or aggressive, either yielding or non-yielding. The research reveals that comparing with other three cases, a qualitative change occurs in the case that the merging behavior and control algorithm are timid and yielding. In this case, the system capacity not only significantly decreases, but also exhibits non-monotonic variation with the AV penetration ratio. The underlying reason is that in Case IV, the timid and yielding strategy forces alternating merging patterns (Type C), which significantly increases average time‑headway and reduces capacity, especially at high AV penetration, which is revealed based on analyzing the vehicle trajectories as well as the vehicle-following pairs downstream of the merge. The finding underscores the pivotal role of merging behavior and control algorithm in shaping the dynamics of mixed traffic flow consisting of HVs and AVs. The results provide critical insights for traffic management, emphasizing the need to account for diverse merging strategies when formulating policies for AV deployment.