Making Intelligent Agents Intelligent

Many simulations environments, particularly those used for training purposes, employ simulated human behavior. To date, users frequently complain that such simulated behavior is not very intelligent and easy to "game" thus reducing the effectiveness of the training. We propose to improve intelligent agent technology through improved techniques for eliciting expert problem solving knowledge on what decisions to make in the simulated environment and improved knowledge representation and reasoning techniques that support intelligent evaluations and plannings in response to user actions within a simulation. Our approach will be demonstrated in a proof-of-concept system that employs intelligent agents in a dismounted infantry fighting environment. This context will allow us to leverage the proposed work with other projects RDC has, developing simulation-based training for the dismounted infantry.