Assessing cyclist safety using infrastructure parameters from OpenStreetMap: The case of Leipzig, Marseille and Edinburgh

Road crashes produce more than one million deaths every year around the world, and they are considered as the eighth cause of death for people of all ages. In addition to loss of life, traffic crashes carry more consequences with high impact on the society, including reduction of quality of life, medical costs, property damage and administrative costs. Cyclists are one of the most vulnerable road users, and there has been an increment on cycling victims in European roads in the last years.
Previous research on road safety has shown the impact of road infrastructure on crash risk and severity, which makes essential to consider it in road safety analyses. Although official traffic authorities have infrastructure inventories, this information is rather difficult to get, and usually it is not suitable for road safety assessments. However, due to the increasing use and contribution of Volunteered Geographic Information (VGI), crowdsourced geographic data is being created around the world, and particularly infrastructure-related data is freely available in OpenStreetMap (OSM).
Motivated by the challenge of reducing cyclist victims from traffic crashes, and the availability of data through VGI, the main objective of this research was assessing cyclist safety considering infrastructure parameters from OSM. Therefore, an inventory of intersections at the city level was obtained, focusing on road infrastructure parameters and cyclist victims. In terms of infrastructure, the intersections were classified by type (number of arms), presence of cycling infrastructure and traffic signals. On the other hand, the cyclist victims were analysed in three different categories: all victims (regardless severity), slightly injured victims, and seriously injured and killed victims.
A methodology to cluster the nodes and ways conforming the traffic network from OSM was presented, taking into account the road hierarchy. This process led to identify the intersections as single points from the clustered nodes, and to count the number of streets arriving to each intersection from the clustered ways. Later, the geolocated victims were matched and assigned to the intersections, leading to conduct a spatial and statistical analysis of cycling victims, including the identification of hotspots. The proposed methodology was implemented in Leipzig (Germany), Marseille (France) and Edinburgh (Great Britain), which were found to be comparable in terms of area and population density, and taking into consideration the crash data availability.
Based on the results by typology, 3-arm intersections presented the highest distribution of intersections with cyclist victims, with at least 50% in each of the three cities. However, comparing the rate of intersections with victims among all intersections of the same type, intersections with five or more arms, and roundabouts had the highest percentages in each city. When studying the different infrastructure parameters, in most cases the results suggested that intersections with traffic signals were safer for cyclists. Similarly, this pattern was also obtained when exploring the different categories associated with the severity level of the victims. Regarding the hotspots analysis, it was found that typically the hotspots were located in non-residential streets, particularly when analysing the most critical severity level (seriously injured and killed cyclists).
Since OSM is a promising data source for replicable road safety assessments, more researchers are encouraged to include it in their analysis worldwide, being aware of data correctness and avail-ability. Additional data from other VGI sources is also advised to complement future analyses.