In the literature automobile driving has almost exclusively been modeled on the basis of the lateral position of the car inside the traffic lane, i.e. lateral vehicle control, and, to a lesser extent, on the basis of the velocity of the car, i.e. longitudinal vehicle control. It has been assumed that the drivers are continuously fully attentive in order to minimize all path deviations. However, this assumption is in contradiction with actual driving, in which drivers normally do not attempt to compensate for all path deviations because, for instance, drivers are confronted with several tasks to perform simultaneously. The models developed for vehicle control are valuable for driving situations in which lateral control requires all the driver's attention, for example during the compensation of heavy sidewind disturbances or during driving a strongly curved road. Normally however, these situations are not frequently met. Therefore, a description of driving should take its starting point differently. In this thesis new approaches are discussed by considering driving within the context of supervisory control. This means that the driver is assumed to supervise the singular tasks like lateral and longitudinal vehicle control, each of which is assumed to be autonomously controlled to a certain extent. The driver only intervenes when the condition of a singular task force him to do so. In this approach an integral description of multitask performance in driving becomes possible.
Abstract