Abstract

The use of traffic conflicts is gaining acceptance as a proactive approach to studying road safety. A traffic conflict involves a chain of events in which at least one of the involved road-users performs some sort of evasive actions to avoid a potential collision. Pedestrian evasive actions are normally manifested by changes in the walking behaviour which is expressed through variations in their speed profile. This paper investigates the automatic detection of pedestrian evasive actions in a computer-vision framework. The study proposes a new measure for detecting pedestrians undertaking evasive actions based on permutation entropy (PE). PE is a robust approach for discovering dynamic characteristics of a time-series. In the current context, it reveals the degree of abnormality in the walking pattern by identifying the deviations from the normal free walking. The methodology is applied and validated using video data from an intersection in Shanghai, China. Results show that the PE-based indicator has a high potential to identify and measure the severity of conflicts that involve pedestrian evasive actions compared to traditional time-proximity measures (e.g. time-to-collision and post-encroachment-time). This research finds many applications in the modern transportation infrastructure monitoring, studying pedestrian crossing behaviour and developing safety programs for vulnerable road-users.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call