Citation

Qayyum A, De Baets B, Van Ackere S, Witlox F, De Tré G, Van de Weghe N. Traffic Injury Prev. 2023; ePub(ePub): ePub.

Copyright

(Copyright © 2023, Informa - Taylor and Francis Group)

DOI

10.1080/15389588.2023.2242993

PMID

37565705

Abstract

OBJECTIVE: Vehicular lane-changing is one of the riskiest driving maneuvers. Since vehicular automation is quickly becoming a reality, it is crucial to be able to identify when such a maneuver can turn into a risky situation. Recently, it has been shown that a qualitative approach: the Point Descriptor Precedence (PDP) representation, is able to do so. Therefore, this study aims to investigate whether the PDP representation can detect hazardous micro movements during lane-changing maneuvers in a situation of structural congestion in the morning and/or evening.

METHOD: The approach involves analyzing a large real-world traffic dataset using the PDP representation and adding safety distance points to distinguish subtle movement patterns.

RESULTS: Based on these subtleties, we label four out of seven and five out of nine lane-change maneuvers as risky during the selected peak and the off-peak traffic hours respectively.

CONCLUSIONS: The results show that the approach can identify risky movement patterns in traffic. The PDP representation can be used to check whether certain adjustments (e.g., changing the maximum speed) have a significant impact on the number of dangerous behaviors, which is important for improving road safety. This approach has practical applications in penalizing traffic violations, improving traffic flow, and providing valuable information for policymakers and transport experts. It can also be used to train autonomous vehicles in risky driving situations.

Language: en

Keywords

road safety; data mining; Pattern recognition; intelligent transportation system; spatiotemporal modeling; vehicle interactions