Understanding and modelling disorderly traffic streams

Abstract

Disorderly traffic streams are those that, simply stated, do not have parallel lines (or lanes) of vehicles but have vehicles distributed more haphazardly in the road space. Vehicles in such streams, while moving longitudinally, change their lateral positions frequently. Their trajectories have a more pronounced wander along the width or the lateral dimension as opposed to those vehicles that primarily move in lanes. This property of disorderly streams dictates that its mathematical models must admit two spatial dimensions (the longitudinal and the lateral). Further, the observed impact of road geometry features like width, curvature, etc., on stream behavior, irrespective of whether the stream is disorderly, also suggests that realistic models of traffic streams must describe the streams using two spatial dimensions. Unfortunately, most of the theories of traffic dynamics are one-dimensional—they only consider the longitudinal dimension. This paper, while describing many of the existing approaches to modelling vehicular traffic behavior builds a case for strengthening two-dimensional modelling approaches that are all, still in their infancy. Given the (1) large increase in computation and data handling capabilities over the last decade and (2) significant strides made in developing tools for observing traffic dynamics at scales and accuracy levels that were previously unimaginable, the authors believe the time has come to develop, calibrate and validate reasonable twodimensional models of traffic dynamics.