Learning objective agent behavior using a data-driven modeling approach

This paper presents a data-driven approach towards the modeling of agent behaviors in a full-fledged, commercial off-the-shelf simulation milieu for tactical military training. The modeling approach employs machine learning to identify behavioral rules and patterns in data. Potential advantages of this approach are that it may improve modeling efficiency and, perhaps more importantly, increase the realism of the training simulator. In this work, we present an architecture outlining the main components of the data-driven behavior modeling approach. Using a prototype that implements the approach, we conduct and present results from an experiment targeting the learning of cooperative military movement tactics. It is shown that the prototype is capable of identifying the rules of the tactics. Moreover, it is shown that the agents are able to generalize such that the learned behavior can be applied in a new setting different from the one observed in the training data.