Congestion tail detection – field test

In a field test, the quality of the congestion tail detections based on floating car data was verified. Existing traffic cameras on previously defined motorway sections served as a reference for the quality check.

Over again, serious rear-end collisions occur on motorways at the tail of traffic jams, resulting in personal injury and property damage. But this does not have to be the case, because timely informing road operators and warning road users have a high potential to improve road safety and traffic flow.

However, a basic prerequisite is that traffic jams and the tails of traffic jams are reliably detected, which is only possible to a limited extent by means of stationary detection by the road operators. A current alternative to congestion warning via stationary detection are services from data providers and navigation service providers, which record traffic disruptions and congestion tailbacks via vehicle-generated data, such as Floating Car Data (FCD), and communicate corresponding warnings to their users (road users).

But what criteria must these vehicle-generated data be subject to with regard to their nature or quality for the purposes of traffic management and traffic information? What are the minimum requirements to be met by the providers of commercially available congestion tail data in the event that the public sector wishes to procure these data as part of a public tender?

These and other questions were addressed in the research project carried out by Rapp in cooperation with the Austrian Institute of Technology (AIT), Vienna and SLR Engineering, Graz, in which we prepared, conducted, analysed and managed an evaluation of commercially available congestion tail data in a field test in real traffic. The aim was to compare the congestion tails recorded by video cameras with the congestion tail reports from four FCD providers in order to draw conclusions about the quality criteria and minimum requirements to be met by the private service providers. A benchmark of the quality of the reports issued by the providers was not carried out. All data were anonymised accordingly.

For the field test, an approx. 1-3 km long motorway section with possibly continuous monitoring by traffic cameras was needed as a test section. A section of motorway 81 between Ludwigsburg Nord and Ludwigsburg Süd in the direction of Stuttgart with a temporary hard shoulder (TSF) system that was not yet in operation proved to be suitable. Using recorded video data, the aim was to determine congestion phenomena with potentially dangerous tail ends and then compare them with the reports from the data providers.

For an automatic video evaluation, an analysis tool was developed that uses special algorithms to record vehicle speeds and vehicle trajectories from the collected video data and filters out potentially dangerous tail ends. The detected tail ends were then either confirmed or rejected visually in a manual video inspection. Finally, within the evaluation platform, the congestion tail messages of the data providers were compared with the analysed video data in order to determine quality parameters of congestion tail data and to draw conclusions on their quality requirements.

However, the results of the comparison showed that the congestion tails in the viewing area of the cameras only partially coincide spatially and temporally with the reports of the data providers. Although a large number of tail ends were identified, these could only be partially confirmed via the reports from the data providers. The limitation of only being able to map a limited section of the real traffic flow with the video analysis meant that the high spatial and temporal dynamics of a congestion tail could not be adequately countered, although this does not mean that the approach of automatic video analysis is fundamentally unsuitable, nor that the tail end reports from the data providers are of poor quality.

Nevertheless, it must be summarised that no concrete values could be derived from the practical test that can be used as minimum requirements for a high quality of congestion tail reports in the context of a public tender. Unfortunately, the results do not allow a clear statement as to whether the data providers can comply with the proposed minimum requirements for the quality criteria, which means that these most important key parameters for specifying the quality of the services to be provided by the contractor are missing for a public tender.

Nevertheless, there are various possibilities to initiate a public tender procedure for the procurement of congestion tail data and, in particular, to have the potential bidders provide proof of the quality of the service to be provided within the framework of the procurement procedure. This could be, for example, a public invitation to tender with a negotiation procedure, in which the bidders have to create a technical concept within the procedure and provide proof with regard to the achievable quality. The findings from the negotiation phase can subsequently serve as input for the performance specification, in particular for the determination of target values of KPIs, requirements for evidence in relation to the award procedure as well as for the provision of the service over the entire contract period. This ensures a high degree of transparency and clarity about the contractually defined services - on the part of the bidders and on the part of the tendering body.

Contact us

Key data

Client: Federal Highway Research Institute (BASt)

Period: 2019 - 2020

Partner: AIT Austrian Institute of Technology, SLR Engineering