The accurate prediction of vehicle travel times is essential to the implementation of many applications within the area of Intelligent Transport Systems (ITS) including route guidance, traffic control, traveller and passenger information, and public transport priority at signalised intersections. Earlier work on predicting vehicle travel times concentrated on statistical techniques. Usually, the vehicle movement is modelled as a stochastic process and some sort of stochastic filtering or smoothing is adopted to predict the vehicle's future travel times. An example of such approach can be found in the route guidance system where link travel times are estimated from probed vehicles. Other examples are presented for predicting the travel times of fleets of buses where each vehicle in the fleet is individually located. In this paper, the movement of vehicles in the traffic network as a dynamic process that has a high degree of determinism that can be resolved by understanding and incorporating the factors that affect the evolution of this process is modelled. In order to achieve such an incorporation of information, a vehicle trajectory estimator algorithm that is referred to as EPTA for Estimation of Plausible Trajectories Algorithm is proposed. The rationale behind this modelling and the details of the implementation of EPTA are first described. Then the results and conclusions of applying EPTA to predict the travel times of simulated buses are presented.
Abstract