As simulation plays an increasingly central role in training events, the requirements imposed upon autonomous entities within those simulations increase. In order to facilitate realistic training experiences, we expect the entities within our simulations to behave in a complex and human-like manner. One aspect of simulating human activities in complex simulated environments centers around enabling the projection of future states of a simulated world, in order to determine the relative merits of potential courses of action. This projection involves introspection of not only the entity's own state, but the state and anticipated actions of other entities within the environment.

This paper presents work sponsored by the Office of Naval Research (ONR) which demonstrates a general framework for how projection could be performed, with a focus on one method of imagining and evaluating future states. A strategy will be explained for how relevant information can be extracted from a simulation environment and converted into an abstract representation of the current state of the world. This representation will in turn be utilized by an abstract qualitative simulator, the NAUTical Intent and Location Utility Simulator (NAUTILUS), which will simulate the course of events by applying Subject Matter Expert (SME) created behaviors. These SME created and validated behaviors also contain a strategy for evaluating these courses of action which will be made use of by the Naval Evaluator of Missions and Objectives (NEMO) in appraising the state of the world. The NAUTILUS and NEMO provide the information necessary for an autonomous entity to determine whether the merit of the intended course of action validates its enactment. This projection framework will be displayed in an example Anti-Submarine Warfare (ASW) training exercise where an autonomous entity will project whether an intended course of action merits execution.