An inclusive framework for automatic safety evaluation of roundabouts

Abstract

This article presents an approach for studying different aspects of traffic safety at roundabouts. An automated safety analysis framework is used to detect different types of traffic conflicts, as well as the inappropriate negotiations and the gap acceptance behavior of drivers. To test the validity of the proposed method, a case study is used for a roundabout in Doha, Qatar. Seven types of traffic conflicts are studied and their severity is identified using the time-to-collision conflict indicator. Four common types of driver inappropriate negotiations behavior are also investigated. The analysis shows that most of the inappropriate negotiations and traffic conflicts are due to drivers' poor lane discipline that can be partially attributed to the poor lane marking. Gap acceptance behavior is also studied by identifying lead, lag, and total gaps. The traffic conflicts, inappropriate negotiations, and gap acceptance results are validated by a comparison with manual observations. The results of the validation process show the viability of the automated approach that produces acceptable results with less time and effort. Moreover, the data collected using this approach provides further insights on roundabout safety evaluation and has the potential for use in the assessment of roundabout design.