Generic agent models for simulations of concepts of operation: Part 2

As detailed in the companion paper (Pritchett, A. R., Feigh, K. M., Kim, S. Y., and Kannan, S. K., "Work Models that Compute to Describe Multiagent Concepts of Operation: Part 1," Journal of Aerospace Information Systems (to be published)), simulation frameworks such as "work models that compute" predict the complex, heterogeneous dynamics of concepts of operation that include physical components, humans, and automated agents. Work is modeled as a separate construct, and agent models are called during run time to execute actions from the work model when they are required. Thus, this paper applies the thought experiment in the form of the following question: What capabilities can be built into a generic agent model that has no internal knowledge of the task or actions to perform? These agent models are "generic" in that they can be called at run time to operate on any work model. One capability in these generic agent models is task management, including prioritizing some actions such that others are interrupted, delayed, and perhaps ultimately forgotten. Further capabilities include anticipation of upcoming actions and maintaining an internal representation of those aspects of the work environment that they act upon. In addition, this paper discusses how a work model may provide several strategies from which each agent may select to adapt their behavior to immediate context.