Cyclist overtaking safety study using video data

While cycling has grown considerably in recent years, it is not as safe as it should be, and the perception of unsafety hinders its further growth. Intersections are the most complex parts of the road network, but cyclists are also particularly vulnerable on road segments when overtaken by motorized vehicles. Laws have been enacted to guarantee safe overtaking events by vehicles with minimum distance and other speed requirements. While there have been several studies on cyclist passing distance using instrumented bikes or vehicles, there have been few attempts at developing an automated system for the safety analysis of cyclist overtaking at a fixed location. This paper presents a computer vision tool to measure the cyclist passing distance and vehicle speeds before, during and after the overtaking. The tool is tested on a case study in Montréal, Canada. It shows that most passing distances are safe, but that most drivers fail to comply with the requirement to decelerate while overtaking cyclists.