For the development of the GIDS system a task analysis of the car driving task is required. This report describes the results of an analysis of the car driving tasks. The task analysis has resulted in a representation of the driver task, a number of decision rules and a control structure for applying the decision rules. The driver tasks are analysed on the manoeuvering level and not on the control level. Together this constitutes an important part of the intelligence to be incorporated in the analyst planner and the small world simulation. It is concluded that the task analysis of McKnight & Adams has a number of disadvantages for our purpose and that a computational approach is preferred. A general model of speed control is described as well as some experimental support for the model. The situations that require a corrective action of the driver are described together with the actions and the way the actions should be executed. These situations consist of a certain constellation of the position of other objects relative to the car, the distance of other objects relative to the car, the velocity of other objects relative to the car, additional properties of other objects of the car and the driver.
Abstract