Research On Early Warning Boundary Of Rear Vehicle’s Avoidance Behavior When The Expressway Leaves And Changes Lanes

The behavior of changing lanes for the purpose of driving out in the exit area of the expressway is called lane-changing behavior when exiting.Rear vehicles in the target lane should slow down during the lane change stage of the vehicle in front to avoid it.Compared with the free lane change of the main line section,the exit traffic flow is complex and belongs to the accident-prone area,and the driver’s intention to change lanes is more urgent.When the driver of the rear vehicle fails to slow down and give way in time,a severe traffic accident may occur.Therefore,this paper takes the process of avoidance of the vehicle behind the target lane when the vehicle in front of the highway exit area exits and changes lanes as the research scenario,takes the vehicle behind the target lane as the research object,and uses the natural driving trajectory data of the exit area to determine the early warning boundary of the rear vehicle in the process of avoidance.The data in this paper includes two sources:(1)NGSIM: US-101;(2)Xi’an Ring Expressway.The expressways in both places are identical in exit form and geometric linearity.After noise reduction and screening matching,3317 sets(NGSIM)and 4474 sets(Xi’an)exited lane change-avoidance data sets were obtained,respectively.Due to the intertwined behavior of the study section,there are also two driving intentions in the avoidance process of the rear vehicle:(1)follow the preceding vehicle into the exit lane;(2)Change lanes to the left and enter the main lane.After the Pearson correlation test,it is found that there are significant differences in the motion characteristics of avoidance vehicles with different driving intentions,so the calculation must be considered independently when constructing the early warning boundary.Secondly,the analysis of motion characteristics and safety evaluation indicators of different driving intentions in the two datasets of China and the United States was carried out,and the following conclusions were obtained:(1)there were differences in the motion characteristics of rear vehicle drivers in China and the United States,but they were basically the same in the selected safety evaluation indicators;(2)There are significant differences in the motion characteristics and safety evaluation indicators of avoidance vehicles with different driving intentions.Therefore,in this study,a support vector machine was used to construct a driving intent recognition model for avoidance vehicles as a prerequisite before the early warning boundary was enabled.In this study,the Xi’an Ring Expressway data was selected to calculate the avoidance early warning boundary,and NGSIM data was selected to test and verify the boundary.Taking the Inversed-Time-to-Collision(ITTC),Margin-to-Collision(MTC)and Time-to-Headway(TH)as the early warning parameters,the Gaussian Mixture Model is used to divide the interaction state of the two vehicles into three types: safety,potential collision risk and emergency.The data with a division probability of less than 0.6 is defined as boundary data,and the boundary data points are used to fit the safety-potential collision risk early warning boundary(general early warning boundary)and potential collision risk-emergency early warning boundary(emergency early warning boundary)respectively.Using NGSIM data for testing,the results show that the avoidance behavior early warning boundary constructed in this paper can make up for the limitations of using a single indicator for early warning,and improve the accuracy of early warning.