The Automated Wingman: an Intelligent Companion for Aircraft Simulators in a Distributed Interactive Simulation

Research in Computer-Generated Forces (CGFs) has developed the ability to instantiate a variety of computer-controlled entities that can operate within a virtual battlespace based on the Distributed Interactive Simulation (DIS) protocols. Unfortunately, with current technology it is often easy to distinguish between human-controlled and computer-controlled entities because of the predictable nature of the computer-controlled entities’ behaviors. This shortcoming may allow training subjects to identify the CGFs, predict their behavior, and defeat them using methods that fail in the real world. As a result, poorly performing CGFs may reinforce improper behaviors on the part of the human trainees and result in ineffective training. Some of the predictability of a CGF is due to an inability to emulate the results of the human decision making process. Our research project, called the Automated Wingman, is aimed rectifying this shortfall by developing aircraft CGFs that exhibit the complex characteristics of human behavior without mimicking the human decision making process. The Automated Wingman is a semi-autonomous entity that incorporates fuzzy logic into its decision making process. Fuzzy logic enables the Automated Wingman to mimic the results of human reasoning by dealing with ambiguity and uncertainty in a way that traditional logic can not. In fuzzy logic, inputs are converted to linguistic variables that embody the uncertainty inherent in real world data and in any decision process. A fuzzy expert system then uses these linguistic variables to determine the appropriate throttle and stick response for the Wingman given the current local situation. The response is defuzzified and presented as inputs to the dynamics model that propagates the aircraft through the environment. This paper describes our current progress toward making realistically behaving computer generated air forces.