This report deals with individual driver behaviour in congested traffic conditions. The aim is to identify the important factors in driving behaviour that have an effect on the occurrence of congestion. In addition, a preliminary model analysis is performed. The aim of this analysis is to demonstrate the potential of such a modelling approach at the level of individual driving. A literature survey is performed in order to establish the most important factors involved in driving in congestion. Furthermore, a model analysis is performed of the car following task, utilising a model of the driver/vehicle combination. The individual driving behaviour in following a lead vehicle could be analysed in detail, for two driving speed (traffic density) levels. The literature review resulted in only 13 interesting articles of which a few focus on car following behaviour and the other ones on specific congestion driving. From the literature review a number of factors could be identified that can play an important role in driving behaviour during congestion. These are: time headway, workload, average speed and standard deviation of speed. The model analysis has shown to provide a useful insight in the various aspects of car following and to predict the quantitative effect of traffic density (driving speed) on car following performance and driver workload. Specifically, car following performance and/or driver workload are more favourable for lower speed traffic. An analysis of car following and collision avoidance performance based on an automatic stop-and-go system shows that the automatic system performance is superior to driver model performance, at the cost of more control activity. From the literature study it can be concluded that drivers have a preferred headway and a preferred speed, but they change this headway (increase) and speed (decrease) in congestion for one or more of the following reasons: 1) In different congestion regimes, drivers adopt a different preferred headway. Only within the same congestion regime individual drivers keep the same headway characteristic, which can again differ among individuals. 2) When they expect lead cars to decelerate, which can be the case in unpredictable high density traffic. 3) When they experience a higher workload, which can be the case in unpredictable high density traffic. The optimal control model approach is adequate to describe the car following task and to investigate individual driving behaviour in congestion. Individual driving behaviour (task performance and workload) can be summarised in terms of only three model parameters. The two performance parameters can be used as inputs for a traffic flow model, yielding a complete model including both the detail of individual driving and the global description of traffic flow dynamic behaviour. (Author/publisher)
Abstract