A hierarchical cooperative merging control strategy for the mixed traffic of CAVs and HDVs

The interactions between vehicles in freeway merging zones can lead to traffic congestion and potential collision risks, resulting in economic loss and environmental pollution. With the development of connected and automated vehicle (CAV) technologies, it is expected to address these issues through trajectory-level vehicular control. However, due to the large number of human-driven vehicles (HDVs) currently in operation, achieving a pure CAV environment will take time. This motivates us to explore the merging control strategies that can deal with a mixed traffic environment involving both CAVs and HDVs. To accomplish this goal, this study proposes a hierarchical cooperative control strategy consisting of a merging sequencing layer and a motion planning layer to facilitate the smooth merging of CAVs in freeway merging zones. First, the globally optimal merging sequence is determined considering traffic efficiency, safety, and driving comfort using the real-time information collected by roadside units. A zero–one integer programming model is built to convert merging sequencing into a shortest-path search problem, enhancing the solving efficiency. Next, a consensus controller with communication delays is proposed considering the state error of all vehicles in the platoon to deal with the future mixed-traffic environment. The local and string stability conditions are derived to establish parameter-setting criteria. Finally, several experiments are conducted to evaluate the performance of the proposed consensus controller and to analyze the impact of CAVs equipped with the proposed controller on traffic flow. The results show that (1) a more reasonable merging sequence can be provided by the proposed algorithm to reduce potential conflicts and help CAVs merge efficiently, and (2) increasing the penetration rates of CAVs can improve the anti-disturbance performance, robustness, and stability of traffic flow in the merging zone. The related algorithms and findings can be adopted in future autonomous driving systems.