Comparison of Cooperative Control Strategies for Freeway Merging Area under Mixed Traffic Flow with Differential Objectives

Along with enhancing traffic efficiency, reducing vehicle carbon emissions has become another primary objective of traffic control, and connected and autonomous electric vehicles (CAEVs) play a key role in this regard. However, in contrast to achieving superior control performance, the differences in control strategies for achieving multi-objectives receive less attention. Combining CAEV-based variable speed limit (VSL) and ramp metering (RM), this study proposes a reinforcement learning-based cooperative control model for reducing total travel time (TTS) and CO₂ emissions (TCE) in a freeway merging area under mixed traffic flow. Results show that TTS and TCE reduction can contribute to each other by relieving congestion. However, it is unable to obtain the optimal improvements in both TTS and TCE simultaneously. A comparison of strategies reveals that maintaining stable mainline flow and low ramp density helps reduce emissions, but this might not be helpful in enhancing traffic efficiency.