A Route-Choice Model for Predicting Pedestrian Behaviour and Violations

Abstract

Pedestrians exhibit diverse behaviours, including crossing violations. Traditionally, development of behavioural models has been divided into route choice and crossing behaviour. Route choice models are stochastic and focused on crowd dynamics, while crossing behaviour models are probabilistic or deterministic and focused on local-level behaviours. Route choice and crossing behaviour are often addressed separately, but they are inherently related. This research proposes a new pedestrian simulation model where pedestrians navigate through an intersection or mid-block environment, modelled as a grid. Each cell is assigned a cost that varies over time based on the presence of nearby vehicle traffic or changes to signal indications. Each pedestrian perceives the costs in the environment uniquely depending on their own personal preferences, like desired crossing gap or comfort committing a violation and seeks to minimize their total path cost. Pedestrians who are more comfortable committing violations perceive a lower cost for committing a violation. This approach integrates crossing behaviour with route choice and models the trade-offs of engaging in a particular behaviour. The proposed model is calibrated using video data. The model was applied to three case-studies: a stop-controlled intersection, mid-block crossing, and two crosswalks along the minor approach of a signalized intersection. The model simulates the trade-offs between walking on different surfaces, as well as the trade-off between waiting for a gap in traffic to cross, versus diverting to the nearest designated crosswalk. In the third case study, the model successfully reproduced the proportion of pedestrians crossing against the signal for the north leg crosswalk but did not reproduce the proportion of violations for the south leg crosswalk, which is across a private access. Further investigation should be undertaken into the causes of this and other differences.