Cyclists’ predictions of what a car driver will do next at intersections

In the Netherlands, 30% of fatal crashes between 2010 and 2015 involved a cyclist [1], with a large portion of these crashes occurring at intersections in urban areas. Contributing factors to driver-cyclist collisions at intersections are not only inadequate visual search, but also incorrect expectations about the other’s intentions [2]. Research also suggests that crashes between drivers and cyclists often happen even when the cyclist must have seen the approaching car [2].The ability to anticipate future events is crucial for safe performance in traffic [3]. Recently, research has started on hazard anticipation in cycling. For example, an experiment using a hazard perception test has found that adult cyclists detect hazards earlier than children [4]. Furthermore, results from an eye-tracking experiment using animated video clips showed that cyclists are more likely to look at an approaching car (e.g., a car on a collision course) than to a car that has stopped before the intersection or a car that has passed the intersection [5]. However, it is unknown at which point in time and based on which visual cues a cyclist can predict that a perceived hazard becomes an actual hazard (i.e., that the car driver will not yield to a cyclist).We developed a video-based survey with the aim to gain an understanding of cyclists’ predictions in hazardous intersection situations. The following research questions were addressed herein:(1) How do cyclists’ predictions of the behavior of a car change in the moments prior to a crash or near miss with that car?(2) Is there a difference in cyclists’ predictions of the car’s behavior between crash and near miss scenarios?Biomechatronics & Human-Machine ControlTransport and Plannin