Wortelen, Bertram and Lüdtke, Andreas and Baumann, Martin
Proceedings of the 22nd Annual Conference on Behavior Representation in Modeling and Simulation
A suitable distribution of attention to task demands is an essential component for efficient handling of multitasking situations. In most cases humans are not consciously aware of how they allocate attention to tasks. Yet they automatically weight their distribution to properties of the task like task value or the frequency of information events for a specific task. The Adaptive Information Expectancy (AIE) model was developed as a dynamic model of attention allocation and integrated into a cognitive architecture. It automatically derives the rate of information events for a task based on the interaction of a formal task model with the environment. The attention of the model is distributed according to these event rates and task priorities. Previous studies demonstrated that a dynamic driver model which uses the AIE model could reproduce many key characteristics of visual attention. In this paper it is shown, how changes in attention distribution are reflected in the task performance of the driver model for the three tasks of (1) keeping the car in the center of the lane, (2) keeping the speed close to 100 km/h and (3) solving a continuous in-vehicle secondary task. Driver model performance is compared to experimental data from a study on human drivers. Shortcomings of the driver model are discussed based on this comparison.
03 / 2013
D3CoS Designing Dynamic Distributed Cooperative Systems