Abstract: Left-turn vehicles will meet many conflict points, which will affect the operation and safety of intersections, so it is necessary to study the characteristics of left-turn trajectories at intersections. It was found that the existing trajectory models did not distinguish well between the existence of conflicting and normal trajectories, which is not helpful for understanding the left-turn characteristics. Therefore, this study recorded the vehicle trajectory of the intersection of Cao’an Highway and North Jiasong Road in Shanghai, China. Firstly, the trajectory reconstruction was carried out by a two-step method, and then the spectral clustering was used to complete the approach road identification. Secondly, the left-turn trajectories were divided into conflicting trajectories and non-conflicting trajectories by using the improved TTC method. Thirdly, the combination of straight line - Euler spiral curve - circular curve was chosen for trajectory modeling, and the influence factors on trajectory were analyzed. The results of multiple linear regression and conflict analysis indicate that, the trajectory entry speed, average speed, minimum speed, entry distance and exit distance (the approach lane selection) on the trajectory have a direct impact on the left-turn trajectory. It was found that the influencing factors of conflicting and non-conflicting trajectories were different, so it is necessary to distinguish and analyze them respectively. In the end, the results provided insight into the characteristics of left-turn trajectories with and without conflicts and provided a basic reference for the operation of autonomous vehicles.
Xuesong Wang*, Ruolin Shi, Weixuan Zhou, Andreas Leich, Xiaoxu Bei. Analysis of Left-turn Vehicle Operation Characteristics at Intersection Using Trajectory Data. Transportation Research Board 101th Annual Meeting, Washington D.C., USA, 2022. 1.9-13.