Head movements predict pedestrian crossing decisions earlier than walking speed

Understanding the behaviour of pedestrians when crossing roads is crucial to improving safety, as pedestrians are the most vulnerable road users, but also the most difficult to predict. Previous research has examined how pedestrians, who intend to cross a road, assess vehicle speed and distance for gap acceptance. However, little attention has been given to the role of head movements as predictor for crossing decisions. Besides by the crossing intention, head movements are also influenced by the approach process to the kerb and the potential traffic directions, but both factors are rarely studied. Here, we compare the predictive power of head movements and walking speed for crossing decisions by examining how pedestrians approach a crossing location while simultaneously looking for crossing opportunities. We used an omnidirectional treadmill combined with a head-mounted display to investigate the crossing behaviour of 36 participants. Our results demonstrate that in the first seconds after pedestrians notice the car, the head-turning frequency is a stronger predictor for the crossing decision than the walking speed. As time passes, the walking speed becomes more useful than the head information and reaches a perfect predictive power shortly before pedestrians cross the kerb. These findings suggest that head orientation could be used as a predictive feature in the initial phase of the crossing and walking speed in the later phase. Ultimately, this research highlights the need for future studies that investigate pedestrian behaviour under more realistic conditions, providing insights relevant to behaviour prediction models for pedestrian protection in automated driving.