Dynamic Safety Estimation of Airport Pick-Up Area Based on Video Trajectory Data
Jiayu Yang, Jaeyoung Lee, Suyi Mao, Junjie Hu
Abstract
An airport pick-up area is one of the most congested transportation facilities. When vehicles go to pick up passengers, they need to frequently change lanes to stop to pick up and leave, resulting in a high collision risk and congestion at pick-up areas. This study aims to investigate safety and efficiency of airport pick-up areas and extract traffic conflicts based on video data. The video data are collected from Changsha Airport in China, and vehicle trajectory data are extracted using Baidu API. The transformation between pixel coordinates and real coordinates is realized through homography transformation, and a GIS tool is used to match the lane data. At the same time, the trajectory is smoothed using Savitzky-Golay filtering. Safety risk indicators: time-to-collision (TTC), modified time-to-collision (MTTC), post-encroachment time (PET), and deceleration rate to avoid collision (DRAC) are utilized to evaluate vehicle collision risks. Modified CUSBoost algorithm is applied for imbalanced trajectory data classification based on four risk indicators and spatial distribution and features of different classes are detected. The results demonstrate that traffic conflicts at pick-up areas are caused by frequent lane changes that lead to overlapping vehicle trajectories and short following distances. It is expected that the findings from this study could help engineers and airport ground transportation managers design and implement effective solutions to enhance safety and efficiency at airport pick-up areas.