Managing critical events: Designing an attention allocation system

Abstract

Future interfaces that intend to support a Commander's decision-making process must be able to recommend prioritized actions such that the decision maker can dynamically focus attention in response to changing environments with uncertain information. A technique using the Choquet integral will be investigated that will monitor the environment by sampling current utility values of criteria and aggregate this information to provide a means to determine the most important information and critical tasks. This paper will present promising initial findings which support that the Choquet integral can be used to represent expert knowledge in the situation where the criteria are dependent. The results will be applied to the ranking of robot controllers.