Understanding Drivers' Anticipation of Taking an Off-Ramp: A Random Parameters Duration Model

Accurately capturing lane-changing behaviour in traffic simulation models is difficult because drivers consider many factors during this task. Therefore, understanding what triggers lane-changing and how lane-change decisions are affected by surrounding conditions aids in improving existing models and/or devising new ones. Existing lane-changing modelling approaches are often predicated on specific rules, imposing strict assumptions that all drivers perform their manoeuvres similarly, neglecting driver heterogeneity. Further, lane-changing is often categorised as mandatory or discretionary based on a distance-based metric, i.e., when further away, it is less critical to perform the lane-change. However, driver decisions to take an off-ramp depend not only on the driver’s characteristics but also on traffic conditions when a discretionary manoeuvre becomes mandatory. This study, for the first time, provides more insights into these contextual aspects by analysing drivers’ lane-changing behaviour in anticipation of taking an off-ramp from a motorway. A driving simulator experiment was designed and conducted to obtain individual drivers’ lane-changing data under three specific driving conditions, whereby traffic volume in the target lane varied from low to high. The remaining time until taking the off-ramp, considering early/late lane-changing starting, is modelled using a random parameters duration modelling approach. This approach captures the panel nature of the data and unobserved heterogeneity. Modelling results indicate that in the high traffic volume scenario, most drivers tend to start their lane-changing manoeuvres earlier compared to a medium traffic volume scenario, reflecting proactive driving behaviour. Conversely, most drivers in a low traffic volume scenario delay their lane-changing manoeuvres, potentially assuming that they will easily find a gap later. Further, age and gender-related impacts are observed and quantified. Findings of this study can aid in enhancing and justifying rule-based lane-changing microscopic simulation models by considering driver heterogeneity and/or traffic conditions in the target lane when approaching off-ramps.