Decision Processes in Simulation-Based Training for ISAF Vehicle Patrols.

摘要

Gaining greater understanding of the requirements or needs of a commander or Operation Analysis Team for modeling human behavior, and the effectiveness of modelling and simulation for training for the Afghan mission, require a modelling process to make the structure of the decision domain explicit. As an initial knowledge elicitation step in this process, we have conducted a pilot study involving two field studies to create an initial model of how a team in a vehicle patrol operates within a training scenario implemented using a virtual training system at the Swedish Armed Forces International Centre (SWEDINT) at the Swedish Life Guard Regiment in Stockholm. The scenarios designed by the instructors at IntUtbE, the pre-deployment training for peace support operations unit within the Life Guard Regiment, are used to train and prepare teams for the real environment. Most importantly, the simulation-based training process creates an opportunity to train a team for specific possible events that can occur in the field, and how to act if those events arise. The method used for modelling cognitive processes involved in this simulated (and real) operational environment began with the Belief, Desire and Intention (BDI) model of human decision-making. BDI has typically been used to model individual motivational factors and cognitive decision processes. Since patrol activities involve groups, the BDI model has been extended with several layers to bring the team aspect into focus. The first field study within the pilot study involved observation of a simulated patrol and recording of the patrol session using voice recording, video, notes taken by an observer, and the screen history of the simulation together with captured eye gaze data upon the screen using an eyetracking system. The eyetracker provides tracking of objects of visual attention of the patrol leader, and dialogue between the patrol leader and driver can be analyzed in relation to these objects.