Analysis of long-term observational data on pedestrian road crossings at unmarked locations

Crossing roads is dangerous for pedestrians. Roads can be crossed at controlled locations, where traffic lights or zebra crossings regulate the behaviour of all traffic participants, or at unmarked locations, where pedestrians typically do not have priority. Technological advances mean observational data on pedestrian road crossing behaviour from public roads can now be recorded almost continuously. Here, we report on such a data collection campaign in Bristol, UK. We record the movement paths of traffic participants within the field of view of commercial camera-based sensors at two unmarked crossing locations. Between January and April 2022, we detect over 30,000 pedestrian road crossings across the two locations. We first explore the time series of hourly crossing counts, finding pronounced and regular temporal patterns that differ between locations. We then investigate the relationship of crossing numbers with road traffic characteristics and extraneous factors, such as university term dates, confirming previous findings on traffic volume reducing crossing frequency and the differences between our study sites. Finally, by studying the timing and distance between consecutive crossings we find evidence for social crossing behaviour, such as groups crossing synchronously. In addition to the specific findings on road crossing behaviour of our study, a key contribution of our work is a case study for how to work with large-volume, low-fidelity observational data on pedestrian behaviour that is becoming increasingly available and has the potential to transform pedestrian road safety research.