Real-Time Queue-End Detection on Freeways with Floating Car Data

Abstract

This paper is a contribution toward an operational use of large floating car data in traffic management. The work focused on a practice-ready approach on highways. The goal was to detect in real time the end of a perturbation. As an entire highway network is not fully equipped with cameras or loop detectors, floating car data have the potential to help detect the end of a moving bottleneck better. This specific zone represents a significant road safety risk. Better real-time detection of the end of congestion is needed. To address this issue, real-world data were analyzed from a French freeway with recurrent congestion patterns. After the quality and the precision of floating car data were discussed, a dynamic spatial segmentation of the network highlighted the relevance of this data source from an operational standpoint. In addition to the empirical network characterization, a systematic detection algorithm able to detect the queue end in real time with a 500-m precision was introduced. Assuming a growing penetration rate of floating car data, the algorithm used only floating car data with simple detection rules and few parameters. The method was validated on real congestion cases. Results proved the accuracy of the detection. The paper discusses the precision of floating car data, and recommendations for road operators are introduced.