Exploring Conflicts on Bicycle Highways Through the SCALE-UP Project: 3D Cameras and Artificial Intelligence in Antwerp Urban Node
On 13 November 2024, the Urban Nodes Platform Event “Moving People and the Economy Across the EU & in Cities” will take place in Antwerp, Belgium.
The province of Antwerp, as part of the SCALE-UP project, is leveraging cutting-edge 3D camera technology and artificial intelligence (AI) to improve road safety at four key intersections along the bicycle highway network.
To ensure privacy, the cameras only retain anonymised footage during the potential conflict, in full compliance with GDPR.
Locations Studied
The four locations studied with 3D camera technology were selected from a list of intersections. That list is the result of extensive data analysis of the bicycle highway network. In this analysis, we combine data about the speed limit of motorized traffic, the road category, visibility at intersections, traffic lights with waiting times and the extent to which the traffic light control ensures that road users cannot conflict with each other. In addition, we also look at accidents involving active road users in the past 3 years. We then estimate the risk of accidents based on these parameters.
The results of the risk analysis have been collected in a dashboard, where you will also find more details about our approach.
Mixed Traffic Intersections:
Both intersections have a similar design, next to a railway crossing and priority for car traffic. An important difference between the two is visibility.
Intersections with Only Bicycle Traffic:
Both locations are on the FR10 (the ring cycle path in Antwerp). The intersections have a different priority arrangement.
Besides collecting objective data to propose improvements for road safety, the camera studies also serve to improve data-driven risk analysis. We want to check whether we have all the relevant parameters to estimate the accident risk of the intersections on the bicycle highway network as accurately as possible.
General Insights
For this research, the Province of Antwerp is using an OTUS3D camera. Discover how it works in the video (Source: Viscando).
Camera research and conflict detection are especially valuable for objectively visualizing changes in conflicts before and after an intervention (change). The comparison of two different intersections is difficult because each location has unique characteristics. For example: slope, curves, visibility, speed and the mix of road users.
Local expertise is vital for accurately interpreting the data.
Current methods for calculating conflicts between cyclists are inadequate, as they rely on the same principles used for car-bicycle interactions, despite the considerable differences in mass and behaviour. Cyclists travel in closer proximity to one another than cars, complicating the assessment of potential conflicts.
Additionally, there is a lack of literature and guidelines to accurately predict accident risks based on conflict numbers and traffic intensity. Data from Boechout and Kalmthout revealed that a higher number of conflicts does not necessarily correlate with more accidents, highlighting the need for a centralised database to consolidate similar studies for more accurate analysis.
The study indicates that additional information, such as traffic intensity, speed (V85), and road gradient, is needed for more precise estimates of accident risks.
Future Developments
The province of Antwerp is also testing camera analysis at a crossroads in Puur-Sint-Amands with a new provider, with results expected by the end of 2024. If you’d like to keep up with insights from this technology, feel free to reach out at: fietsen@provincieantwerpen.be.
Authors: Kim Verbeeck, Sara Van Elsacker (Provincie Antwerpen)