Assessing the safety of cycle paths with artificial intelligence

04 May 2021 - Published by Matteo Candelari

This article was originally published in Dutch by Fietsberaad/CROW and can be found here.

In the Province of North Brabant, an instrumented bicycle will be used to assess traffic safety. This is an initiative by the Province, Royal HaskoningDHV and BAM Infra Nederland.

The Province wants to gain insight into which bicycle paths in the province may be made safer. According to the province, until now it has been difficult to gain a quick, simple and unambiguous insight into the safety of the cycle path, because the data is either outdated or not available (anymore). With this pilot project, the parties involved are trying to develop an inventory method that can easily provide a good insight into the safety of the cycle path. In the future, this may also help other provinces and municipalities to improve the traffic safety of cycling paths.

For the pilot project, both bicycle and rider will be fitted with sensors: LiDAR (Light Detection And Ranging), GPS and cameras. This equipment scans the surroundings of the bicycle and processes these in a digital file. The parties analyse the collected data with the help of algorithms and artificial intelligence models to accurately map out the cycle path. Thus, they expect to get a good picture of the bicycle path, the obstacles and other physical matters that are important for traffic safety.

Early this year, the first test took place on the F59. This route was chosen because of the different types of road surface, priority situations and design choices that occur here and because it has been previously analysed in different ways. This makes the results of this inventory method easy to interpret and compare with known data and insights.

© Fietsberaad/CROW