Developing a Cognitive Model of Expert Performance for Ship Navigation Maneuvers in an Intelligent Tutoring System

The goal of this project is to develop a cognitive model of expert ship-handling performance. This model was integrated with an intelligent tutoring system and an immersive visual simulation used by the U.S. Navy. This intelligent tutor and expert cognitive model (written in a Java-based version of ACT-R) provides feedback to the student based on the student actions in order to reduce workload on the instructors. The nature of ship navigation and the requirements for the intelligent tutor presented unique challenges for development. This paper describes how the resulting cognitive model balances a need for expert performance while compensating for student error, uses perceptual heuristics when the ACT-R vision module is not feasible, and how these and other issues affected model development. Future plans for system test and evaluation are also discussed in the context of improving training.