Identifying factors that contribute to collision avoidance behaviours while walking in a natural environment

Abstract

Busy walking paths, like in a park, a sidewalk in a city centre, or a shopping mall, frequently necessitate collision avoidance behaviour. Lab-based research has shown how a variety of situation-specific factors (e.g., distraction, object/pedestrian proximity) and person-specific factors (e.g., pedestrian size, age), typically studied independently, affect avoidance behaviour. What happens in the real world is unclear. Thus, we filmed unscripted pedestrian walking behaviours on a busy [~]3.5 m urban path adjacent to the water. We leveraged deep learning algorithms to identify and extract walking trajectories of pedestrians and had unbiased raters characterize interaction details. Here we analyzed over 500 situations where two pedestrians approached each other from opposite ends (i.e., one-on-one pedestrian interactions). We found that smaller medial-lateral distance between approaching pedestrians and a lower number of surrounding pedestrians (i.e., smaller crowd size) predicted an increase in the likelihood of a subsequent path deviation. Furthermore, we found that whether a pedestrian looked distracted or held, pushed, or pulled something while walking predicted the medial-lateral distance between pedestrians at the time of crossing. Although pedestrians maintained a larger personal space boundary compared to lab settings, this is likely because of the outdoor paths width. Overall, our results suggest that collision avoidance behaviours in lab and real-world environments share similarities and offer insights relevant to developing more accurate computational models for realistic pedestrian movement.