Freeway weaving phenomena observed during congested traffic

Abstract

This article focuses on modelling the acceleration–deceleration behaviour of drivers during freeway weaving manoeuvres, under heavy traffic conditions. The term ‘heavy traffic conditions’ refers to the period which is fully congested, i.e. queues reside on both freeway and ramp approaches and there is no exogenous flow restriction from downstream of the weaving section (active weaving bottleneck). Data capturing a wide range of microscopic information were collected using video-recording and image processing techniques. The data was collected at two weaving sections on urban freeways in Tokyo and Melbourne. These data provided the fundamental information for investigating the drivers weaving behaviour. A theoretical framework for modelling weaving driver acceleration–deceleration behaviour is presented. The framework uses a stimuli–response psychophysical concept as in its basic formulation. The collected data are used to calibrate the proposed model. A particular feature of the proposed model is that it captures the acceleration–deceleration behaviour of drivers during the entire process of weaving manoeuvre. The results indicate that on average a 0.72 s time gap exists before weaving drivers respond to stimuli. It is found that the surrounding freeway vehicles significantly affect the weaving vehicle acceleration behaviour. Finally, the proposed model was implemented in a traffic simulation and tested using independent data to demonstrate the capability of the proposed model.