Strata (Synthetic Teammates for Realtime Anywhere Training and Assessment): An Integration of Cognitive Models and Virtual Environments for Mobile Scenario Based Training

Abstract

Through an integration of synthetic agents, intelligent tutoring, and scenario-based training in a simulation environment, Synthetic Teammates for Real-time Anywhere Training and Assessment (STRATA) is being developed to overcome conventional training limitations by providing deployable training that offers on-demand practice for individuals and teams, with or without instructors, and without requiring users to be co-located. STRATA will use synthetic agents embedded in a virtual training environment when human players needed to play the part of team members are missing. As part of DARWARS, DARPA's Training Superiority program that is establishing persistent, online training worlds, STRATA will provide these training opportunities with low-footprint, portable hardware. An Event-Based Approach to Training (EBAT) is being used to facilitate highly context specific training and to support generalization to a variety of mission types. The first STRATA application demonstrates Close Air Support (CAS) training. A cognitive task analysis of representative CAS missions, identifying decision making requirements, radio communications, and coordination, has been conducted in order to develop our synthetic models and help identify observable triggering conditions necessary to measure performance targeted in the EBAT method. In order to support the project goals of individual and team-centered training, a generalizable training architecture, and the reuse of cognitive agents across mission types, scenario and training management systems, as well as a cognitive agent library (CAL) are being developed.