A GPS Study of Car Following Theory

Abstract

Car following theory is a small but important element of traffic modelling, which has been researched in various forms for over 50 years. It's description of a theoretical interaction of 2 or more cars while driving is an essential input for transport network models as it directly implies the assumed space and time they will take up while driving. This in turn will affect the theoretical capacity of roads in such models. It is therefore obviously desirable to make this assumed interaction as close to reality as possible.

Car following relationships have traditionally been expressed as mathematical or empirical models, compatible with larger scale transport modelling programs. Although several of these models have been proposed, many of them have proved to be inaccurate or inappropriate for urban specific flow conditions due to the fact that they were calibrated with data taken from freeway environments. Considering that most traffic modelling is undertaken to describe urban flow conditions, this is a serious shortcoming.

This paper provides a summary of a recent study I undertook for my undergraduate honours thesis project at UNSW under the supervision Dr. Peter Hidas. The objective of the project was to address the lack of accurate urban specific car following data by means of an unprecedented application of differential GPS to this task.

As well as negating the traditional problems of data inaccuracy and manipulation, this project also provides a number of valuable conclusions about car following in practice, and the strong potential for further advances in the field of traffic modelling.