Culturally Aware Agents for Training Environments (CAATE): Phase I Final Report

摘要

Recently, the U.S. Army has conducted a wide range of missions within the context of very different cultures and languages. These missions often require junior leaders and Soldiers to interact and communicate effectively with people whose cultures, languages, lifestyles, and beliefs are very different from those found in the U.S. Computer-based training in virtual environments has the potential to train Soldiers to rehearse missions with a sound knowledge of the relevant local cultural context. Existing computer simulations of culturally situated agents representing humans are currently very limited in fidelity, making them unsuitable for training and rehearsal. This effort investigated, designed and demonstrated the feasibility of a two- step approach addressing the modeling of believable cultural agents. First, a mission essential competencies approach identifies key skills needed in training. Second, a modeling toolkit for designing computer-controlled agents for cultural training applications was described. The approach uses social network modeling technologies to develop models of interconnected agents within a graphical environment and a human behavior modeling tool for simulated agents based on the cultural context. This approach was demonstrated by developing an integrated prototype that dynamically created cultural behavior in a virtual environment.