Automated Behavior-Based Interaction Customization for Military Command and Control

In this paper, we describe a proposed architecture for behavior-based customization of user interaction with an adjustably autonomous system. Our approach will use a GOMS-based task representation and Bayesian plan recognition to determine the user’s active tasks. The active task set will be modeled using a cognitive architecture to assess cognitive, perceptual, and manual workload. The resulting user assessment will then be used to determine what, where, when, why, and how the autonomous system can assist the user, adjust its level of autonomy, or perform other customizations to improve the user’s overall performance. The system will be developed in the context of military command and control. The design of this architecture raises a number of scientific and engineering challenges, which we will also discuss.