Modeling Aviation Crew Interactions with Automation Using Cognitive Task Analysis (original) (raw)

Abstract

ABSTRACT Aviation automation has become increasingly complex, multi-functional, and autonomous. These changes have the potential for both performance increases as well as unanticipated and undesirable side effects such as erosion of motor or cognitive skills, novel types of errors, and so forth. Specific understanding of both the positive and negative effects is critical for reliable and valid measurement of crew automation performance, designing and refining appropriate crew automation training, and redesigning or altering future software and hardware for cockpit automation systems. One method for achieving the required specific understanding of how crews interact with aircraft automation is the use of cognitive task analysis. Specifically, this paper will describe the use of one cognitive task analysis technique (NGOMSL) applied to modeling crew activities that take place during the climb or descent phase of flights. Natural-language GOMS (Goals, Operators, Methods, and Selection Rules) was developed by Kieras (1997) based on Card, Moran and Newell's (1983) conceptualization of task decomposition into a hierarchical set of goals, which can be operated on using a set of methods and selection rules. The technique allows the researcher to specify the alternative set(s) of actions that must be carried out in order to achieve a set goal (e.g., the activities that a pilot must carry out to bring a plane down from one specific altitude to another). The usefulness of this technique to the development of a computational cognitive model has been explored for the tasks of making altitude, heading, or airspeed changes based on Air Traffic Control commands during realistic flight scenarios during the climb or descent phases of flight. The task analysis specifically focused on the cognitive demands on the pilot responsible for interacting with the automation during these phases of flight. The task analysis included relevant details of the automation interface such as the panels used for input and output displays, as well as relevant cognitive processes such as perception, understanding, memory recall, evaluation, and decision-making. This information was combined with eye tracking data collected from pilots interacting with a low-fidelity simulator. These data informed our design decisions about what information pilots are acquiring from the flight deck while working with automated systems during climb or descent. The information gained from these sources is now being integrated into a model of crew interactions with the automation and with each other. The presentation will focus on the usefulness of cognitive task analyses in understanding how automation affects the performance of both individual pilots and crews.

Jeff Hansberger hasn't uploaded this paper.

Let Jeff know you want this paper to be uploaded.

Ask for this paper to be uploaded.