A novel methodology to monitor passenger mobility performance in urban subway stations

Abstract

Sustainable smart technology is important to prevent roaming passengers and improve pedestrian space environment by providing effective route information service in subway stations. A pedestrian trajectory collection system based on LiDAR sensors can continuously track individual passengers with a wide detection range and high accuracy for monitoring pedestrian space in a subway station. This study developed a methodology for detecting and classifying abnormal passenger trajectory patterns in a subway station using passenger trajectory data collected from LiDAR sensors. Further, feature vectors were extracted toward the reliable characterization of passenger walking trajectories. An ensemble method that combines decision tree, support vector machine, and artificial neural network was adopted to increase the classification accuracy for abnormal passenger trajectory patterns. The proposed ensemble method was able to obtain 94.4% classification accuracy, which was superior to that of a single classifier. In addition, this study devised a performance measure to identify the performance of passenger mobility in subway stations, referred to as passenger mobility index (PMI), based on the classification results. The proposed methodology is expected to be utilized to provide a passenger mobility evaluation system based on intelligent facilities such as smart construction and safety.