Journal of Korean Society of Transportation ISSN:1229-1366(Print) 2234-4217(Online) 대한교통학회지

The effective provision of route information in subway stations is essential for not only the edestrian guidance but also the evacuation in emergency situations including natural disaster and terrors. This study developed a methodology for characterizing the pedestrian walking trajectory patterns, which can potentially represent the performance of information provision and facility operations in subway stations. Pedestrian tracking data was collected at the Samsung subway station using a LiDAR-based tracking technique. Pedestrian trajectories were classified into four groups such as pacing, lapping, stay, and inefficiency. Feature vectors were extracted toward the reliable characterization of pedestrian walking trajectories. A heuristic classification algorithm based on extracted features was then developed for the trajectory labeling. Approximately, 20% of pedestrians in our dataset were identified as having abnormal trajectories. The outcome of this study is expected to be useful in analyzing pedestrian trajectory patterns.