An example of variable levels of detail for object-based decomposition in J-MASS

Abstract

In an object-based modeling system such as the Joint Modeling and Simulation System (J-MASS), software objects are selected to represent items in the real world. An item model reflects, at implementation, compromises among the realism needed to reach a simulation goal, the data and algorithms available to capture desired behaviors in software. And the desired throughput of the simulations in which the item model will run. Other weighting factors for design choices are possible based on the client, the budget, and the development schedule. This paper describes the object decomposition of the human operator portion of some large complex system, perhaps an aircraft, a nuclear power plant, or a subway train. By assuming that the human operator has both an active role (follows a plan) and a reactive role (responds to cues from the workplace), and also that operator responses to the external world are predominantly indirect (sending a message, pressing a switch, setting a new number), one can identify and map out some of the objects found in the operator part of a human-managed complex machine. The main focus of the discussion is on the process and criteria for selecting objects at several levels of detail based on simulation goals and costs. The "penalties" incurred by choosing a coarse level of detail as opposed to a fine level of detail are noted. The paper also addresses some basic issues, including perception versus ground truth, repetitive tasks versus interruptions, and handling of human factor phenomena such as task saturation and fatigue.