Cognitive Performance Modeling for Cockpit Devices and Tasks
Just as computational models that simulate air flow over an airplane wing allow more rapid and cost-effective exploration of different wing designs than building and testing physical prototypes, computational cognitive performance models allow more rapid and cost-effective exploration of different cockpit device designs than building and testing functional prototypes with pilots. Computational cognitive performance models (CPMs) have a 25-year history of success in the domain of desktop computer applications; the purpose of this 5-year research project is to bring such models into domains of interest to Boeing.
There are several critical differences between CPMs for desktop computer applications and the cockpit, which have been highlighted in videotapes of pilots doing cockpit tasks provided by Boeing in the fall of 2007. This research project expects to make progress on two of them, as follows.
1. Tasks on desktop computer applications use a very limited set of input devices, the use of which has been previously modeled in CPMs, i.e., typing and pointing and clicking with a mouse. Cockpit tasks require a different set of input devices and physical actions, e.g., reaching in 3-dimensions to an overhead device, grasping and turning a knob. Although there has been substantial data collected in the last century about such actions (e.g., by the Motion Time Measurement, MTM, community), these data have not yet been integrated with existing CPMs.
2. Many desktop computer applications are user-paced, that is, the computer does something only in response to an action by the user. Therefore, CPMs in the desktop domain have been able to make useful predictions under this simplifying condition. Desktop applications are starting to break this condition (e.g, AI systems that you set up to interrupt you when an important event happens that the AI system has been monitoring), cockpit applications are almost never as restricted. Forces and agents external to the pilot’s control, including weather, mechanical systems, Air Traffic Control, other airplanes, etc., often impinge on the tasks of the pilot. This necessitates a change in the CPM modeling process. Not only does an analyst have to model specific devices and tasks, but he or she has to specify scenarios of use that include the actions of such outside forces and agents (e.g., a scenario in which a fire alarm occurs).