Search

Through the use of AI technology, the potential exists to make revolutionary changes in military command and control. Knowledge based systems can enable a commander and staff to accommodate large amounts of information efficiently and effectively, to generate operations plans rapidly, and thereby allow exploitation of early maneuver opportunities. This paper addresses the issues of application domain and human factors design practice as they relate to the development of effective user system interfaces for expert systems.

A software development environment for prototyping, simulating, and evaluating advanced concepts in US Army battlefield information systems was designed and implemented as a lower cost alternative to complex development and simulation facilities. The workstation-based development environment was designed to evaluate a variety of soldier-machine interface design requirements and specifications, as well as communications and database concepts for future Army vehicle and weapons systems. Using this environment, an armor maneuver control application was implemented, demonstrated, and evaluated in a networked testbed. The application demonstrated the creation and transmission of tactical graphic and text information, automatic message routing, and rapid reconfiguration of the soldier-machine interface. Use of the development environment has demonstrated that battlefield information system prototypes can be quickly and inexpensively simulated and evaluated, and can be used as an adjunct or alternative to other simulation facilities.

Work time estimation techniques, as an alternative to direct work time measurement methods, were replicated using preventive maintenance personnel as expert estimators. Two groups of maintainers, those familiar with maintenance activities on specific equipments included in the study, and those with generic preventive maintenance experience, estimated subtask completion times on electronic equipment. Both groups' task completion time estimates correlated very highly with observed task completion times, indicating that general knowledge of maintenance is sufficient to produce strong prediction models. The relationship between generic experts' estimates and observed task completion times was further strengthened by providing more detailed subtask descriptions during the estimation process.