Intelligent adaptive training technologies augment or emulate the role of human instructors to support self-training through experiential practice, with individualized guidance and feedback. They can be powerful training tools, especially as budget constraints trigger reductions in instructional manpower and in the opportunities for live training. But the complexity and diversity of learning objectives and practice environments can make the authoring task time-intensive, frequently requiring a highly customized development process. There are no universal authoring solutions, although concepts have been proposed to reduce the burden by isolating reusable components. For example, an abstracted simulation integration layer can standardize performance data collection, so that instructional components and even logic can be reused across applications. In practice, an integrated trainer draws a virtual dividing line where instructional logic is either more or less abstract from the state and event data flow in a simulation environment.

This paper describes a self-training capability integrated with an existing tactical decision making game, as a case study for the tradeoffs in abstracting a set of automated assessment measures and techniques for complex real-world training objectives. The game environment is Follow Me, a small unit leader tactical training game used by instructors and cadets at the United States Military Academy at West Point. This training audience presents a classic use case for self-directed training; due to limited classroom time, cadets previously received little direct feedback on their own exercise performance. However with the addition of automated evaluation and feedback capabilities, the reach of instructors is extended and replicated. While this trainer operates with a two dimensional game, some of the same performance measures could be applied to a three dimensional game or other simulation platform. Examples from the Follow Me trainer are explored in terms of generalization and the entailed authoring implications for both instructional components and experiential environments.