Work models that compute to describe multiagent concepts of operation: Part 1

This paper describes a simulation framework and engine created to address several concerns in the design of multiagent concepts of operation, focused on the construct of "work." This framework incorporates the capabilities of several approaches, including cognitive work analysis, system-level modeling, and agent-based simulation. The framework promotes consistent, systematic models of the collective work performed by any number of agents within a concept of operation. As a side benefit, the ability to simulate work models also provides a mechanism to validate that the description of a concept of operation provides a complete and accurate representation of the agents' collective activities and of the work environment they act upon. The framework defines several computational structures: "actions" describing work processes, "resources" describing the state of the environment, "decision actions" for strategy selection, and "functions" allowing compositional modeling at multiple levels of abstractions. Timing constructs ensure correct temporal semantics, even with a wide variety of computational models, including discrete event forms and integration of continuous dynamics. Actions can be allocated during run time to agents that then execute them. Representative results from a study of function allocation between a flight crew and an autoflight system illustrate the insight provided by these models and their simulation.