This paper describes the Computer Generated Forces (CGF) behavior development on the Air Force Research Laboratory Security Forces Distributed Mission Training (SecForDMT) technology development program. The near-term goal of this program is to develop distributed training technology that will allow Air Force Security Force decision-makers to practice the planning and execution of air base defense. The system must be distributed because Air Force Security Forces are drawn from many separate home bases, and receive only limited training as a team before deployment to a contingency site. The training target is the decision-makers (e.g. Squad Leaders, Flight Leaders, S-3, Base Defense Commander) as opposed to the trigger pullers in the fire teams. This emphasis on decision-makers is due to the fact that needs assessments have indicated training requirements for these positions. We will use CGFs to simulate fire team members, other friendlies, neutrals and threats in order to generate situations requiring decisions by the trainees.

The CGFs being developed on this project are different from other CGFs in a number of ways. These CGFs do not exhibit cold war behavior of always fighting to the death. Although opposing force (OPFOR) CGFs are capable of lethal force, they do not automatically shoot given intervisibilty. Neutral CGFs are capable of a range of behaviors varying from peaceful intent to lethal threat. Security Forces CGFs who observe such threats issue doctrinally correct situation reports (SITREPs) for students' situational assessment, decision making, and operations order/fragmentary order (OPORD/FRAGO) formulation. Security forces CGFs challenge, engage, or capture threat CGFs in accordance with the rules of engagement (ROE) stated in the trainee-generated OPORD as well as level of compliance by OPFOR CGFs. Since these CGFs will be controlled directly by the trainees instead of an experienced CGF master, a simple, understandable user interface is a major design requirement. This paper discusses the process of defining behaviors for computer generated forces and how these behaviors were implemented. It also discusses initial evaluations of behaviors, lessons learned and future plans for extending and improving these behaviors.