A simulation model for predicting driver behaviour and system performance when the automobile driver performs concurrent steering and auxiliary in-vehicle tasks is described. This model is an integration of two previously existing computerized models referred to as the `procedural model' and the `driver/vehicle model.' The procedural component deals primarily with in-vehicle tasks and with the task-selection and attention-allocation procedures, whereas the driver/vehicle component predicts closed-loop continuous control (steering) behaviour. Given descriptions of the driving environment and of driver information-processing limitations, the resulting integrated model allows one to predict a variety of performance measures for typical scenarios. These measures include time histories for vehicle state variables such as lane position and steering wheel deflection as well as allocation of visual and cognitive attention. Model calibration and validation are discussed, and use of the model in analysing complex task situations and in generating human factors guidelines is demonstrated.
Abstract