Abstract

During the reconstruction of an accident, it was necessary to illustrate a lane change process of a cyclist in order to understand the course of the accident and to work out whether this accident was avoidable. For this purpose, the duration, length, and trajectory of the cyclist's lane change process were of particular importance but barely known. So far, only the lane change processes of trucks, cars, and motorcycles have been thoroughly investigated and evaluated in current literature. Therefore a test series was performed to figure out important parameters for lane changing processes of cyclists.

In order to identify the characteristics of a lane change process of cyclists, driving tests with a total of almost 15 volunteers and different bicycles were carried out. A lane with a width of 3.5 m was available to the participants. They were instructed to make one normal and one sudden lane change. In addition, the volunteers drove a lane change procedure and marked their change of direction by hand signal. The driving procedures were filmed with a camera in the direction of travel and from a steady drone position from above. In addition, defined points were marked on the roadway in order to make subsequent evaluation possible.

The video recordings from the drone were time-stamped, split into frames, and the beginning and end of each lane change procedure was defined. The individual frames were scaled computer-aided using the defined markers on the roadway. Thus, the length, duration, and lateral offset of the lane change have been measured or calculated. In addition, the average speeds were instantly available. Using the drone recordings, it was also possible to mathematically characterise the driving trajectories.

In order to reflect the perspective of a motorist behind the cyclists, video recordings were also taken in the direction of travel (cyclists seen from behind). Using a similar approach to the processing of the video footage as with the drone recordings, the aim was to evaluate at which point the initiated lane change process is visible to the following road users.

Due to our own series of experiments, it was possible to identify the trajectory, the average duration, and the length of lane changes of cyclists at a lateral offset of 2 to 3 meters. In addition, we were able to determine a guidance value at which transverse offset the lane change can be detected at the latest for road users behind a cyclist. Based on our own tests, we can provide important parameters for the representation of lane change manoeuvres of cyclists in a distance-time context for future accident analyses.