Modeling Vehicle Interactions during Freeway Ramp Merging in Congested Weaving Section

摘要

The difficulty to capture the interactions between vehicles in congested merging area during lanechange process hinders the development of microscopic traffic modeling. The main goal of thispaper is to quantify the interactions between vehicles during lane change in congested weavingsection and try to incorporate them into vehicles’ (merging vehicles, their putative leaders (PL)and putative followers (PF)) acceleration-deceleration models. Based on the findings byanalyzing the US101 data, the yielding behavior of merging vehicles’ putative leader (PL) formerging cooperation and the lateral separation between vehicles in weaving section (bothfeatures largely ignored in the past research) are introduced into our proposed accelerationdecelerationmodels. The visual angle information is employed as the stimuli in our models topresent the effect of lateral movement of vehicles. The particular car following behaviors areincorporated in the modeling, such as the PF gradually changing its car-following leader from PLto merging vehicle, which depends on the relative locations of vehicles.The model calibration and validation results based on field data have demonstrated thatthe proposed acceleration-deceleration models qualitatively simulated the driving behavior ofvehicles in the lane change process and obtained acceptable training and testing errors. To verifythe cooperation behavior of PL in lane changing, a comparison result of the proposed PLacceleration-deceleration model and a base model implied that incorporating the effects ofmerging vehicle on PL into the models could enhance the realism of the lane changing model.Findings from this study could contribute to the understanding of interactions between vehiclesduring complex lane-changing behavior.