| Popis: |
Vehicles often face congestion at airport curbsides because private vehicles, taxis, and ride-hailing vehicles compete for limited space while dropping off passengers. This competition may lead to blockages and long dwell times, thus worsening the congestion. To solve this issue, management strategies such as first-in-first-out queuing or fines for excessive dwell times have been suggested. However, there is a lack of reliable video data and analysis of drop-off behavior at airport curbsides. In this study, we empirically analyzed the key characteristics of passenger drop-off behavior at the Airport T2 terminal in Guangxi Province, China, which handles approximately 18 million passengers per year. First, we extracted the relevant features of both passengers and vehicles by deep learning algorithm, such as the direction of passenger movement, vehicle type, vehicle location, trunk state, and passenger drop-off time. Subsequently, we constructed new features, such as driver behavior and passenger behavioral complexity, based on the original features. We used least-squares regression and logistic regression to analyze the data. Our analysis reveals that the drop-off time of passengers primarily depends on the complexity of their behavior during the drop-off process. Additionally, we observed that specific features, such as driver behavior and vehicle type, could be employed to estimate passenger drop-off behavior. These findings have practical implications in providing valuable insights into the design and management of airport curbside areas and future strategies for connected vehicles. |
