Supervisory Behaviour In Automobile Driving: New Approaches In Modelling Vehicle Control

TWO METHODS FOR MODELLING driving performance as a supervisory task are presented. The first is based on a preview-predictor model, which is used to predict how long a driver can neglect lateral position errors and thus behave in a supervisory, open loop mode. On the basis of this approach the so-called Time-to-line-Crossing (TLC) concept was developed. At each moment TLC represents the spare time before the automobile will reach either edge of the lane, assuming a fixed steering strategy.

The second way of modelling driving is based on the optimal control model (OCM). This model predicts vehicle control based on the driver's internal estimations of perceptual variables like lateral position. Lateral speed, etc. The increase in the uncertainty of these internal estimations during the temporary absence of the perceptual variables is used to predict driver's spare time.

The value of both models was tested in an experiment in which subjects drove under conditions with temporary occlusion of visual input at different speed levels. Subjects' self-chosen occlusion times could be explained very well by both models. Suggestions are given as to how these models can be applied to evaluate vehicle characteristics as well as road geometries.