ESTIMATING THE DRIVER-YIELDING BEHAVIOUR AT UNSIGNALIZED CROSSWALKS

The unpredictable nature of vehicle response at unsignalized crosswalks is a considerable problem in ensuring pedestrian safety. The study aims to enhance pedestrian safety by analysing driver-yielding behaviour at unsignalized crosswalks, focusing on pedestrian attributes and oncoming vehicle categories. The research examines driver-yielding rates influenced by vehicle speed and pedestrian characteristics, utilising data from two-lane two-way divided roads with separate estimates for near and far lanes. The experiment utilizes drone video footage to capture the vehicle trajectories while still camera images to analyze pedestrian behaviour at the median or further lanes when crossing the first two lanes simultaneously. Driver-yielding rates were accurately captured using Tracking Software, allowing for precise analysis of variations. The trajectory diagrams developed were clustered into different vehicle categories and results show that HCVs showcase the highest speed drop at 48.28% while motorbikes exhibit a comparatively lower decrease at 32.2%. This driver-yielding behaviour was modelled using binomial regression. Vehicle category and position, pedestrian position, and pedestrian demographic characteristics were used as explanatory variables for the model prediction. This research analyses driver-yielding behaviour and vehicle response dynamics to understand key factors influencing pedestrian safety at unsignalized crosswalks, enabling targeted interventions and traffic management strategies.