The generation of synthetic O-D demands from real-time vehicle probe data: Potential and limitations

Abstract

The provision of two-way communications between the central traffic control center and all vehicles, which are equipped with route guidance systems (RGS), makes it possible to assemble very efficiently a continuous estimate of the prevailing dynamic or time-varying O-D (Origin-Destination) matrix. This paper describes the mathematics of a four step procedure for developing such a time varying OD matrix from vehicle probe data and discusses the important statistical limitations of the final OD estimates. It is shown that, unless there is a very high level of market penetration in the network, the resulting OD estimate will contain such a high level of statistical uncertainty as to render it of limited practical value. consequently, it is anticipated that the resulting OD estimate will even in the long term not be unsuitable to be utilized on its own, and that instead this estimate will always need to be further processed as a seed solution matrix for a subsequent dynamic synthetic OD generation analysis.