Examination of pilot benefits from cognitive assistance for single-pilot general aviation operations
Sara A. Wilkins
DOI: https://doi.org/10.1109/dasc.2017.8101987
2017-09-01
Abstract:Evaluating the human factors of new cockpit technologies is a time consuming and expensive task. A typical human-in-the-loop evaluation can take several months to design, execute, and document. Moreover, retaining statistical power requires placing hard limits on the technological variations and scenarios to be tested, drawing out the iterative design process. These studies, however, ensure that flight deck technologies are efficient, effective, and, from a pilot mental effort perspective, manageable. While eliminating human-in-the-loop evaluation is neither recommended nor desired, methods that can reduce the number of design cycles and help human-in-the-loop evaluations target the most promising design concepts are valuable. One such method is fast-time simulation of the pilot activities to assess task times and working memory load using human performance modeling. Well-established task analysis methodologies, i.e., Goals, Operators, Methods, and Selection Rules (GOMS), with theories of working memory and mental effort are able to provide estimates of temporal and mental effort for a task [1]. The MITRE Corporation developed a tool for this purpose, which is a cognitive calculator or “Cogulator.” This paper describes an example of using Cogulator for modeling mental effort with and without a pilot-oriented cognitive assistant, called the Digital Copilot. The Digital Copilot, developed through sponsorship from the Federal Aviation Administration by The MITRE Corporation, is a working prototype that provides contextual information and reminders to the pilot in a timely manner [2]. This study compares two different modes of operation using Cogulator: a single pilot's thoughts and actions during the approach phase of flight without the Digital Copilot and another with the Digital Copilot. Cogulator uses the task decompositions in each mode of operation to estimate three metrics: the pilot's task times, heads down times, and working memory load. Task time is a summary statistic that describes the amount of time that elapsed for a task to be completed. Heads down time is the amount of time that a pilot spends looking at information within the cockpit instead of outside. Working memory load is the cognitive construct in which information is temporarily stored and manipulated to complete complex tasks [3]. The paper describes Cogulator, the evaluation method, and the results of the comparison. Results are presented in terms of task completion time, heads down time, and working memory load during the approach phase of flight for the following five tasks: check automatic terminal information service (ATIS) frequency, receive contextual frequency, review and follow a checklist, determine if the tower is open, and determine the preferred runway. Of these five tasks modeled, results show that the Digital Copilot provides time savings in all tasks except for check ATIS frequency; heads Down time savings for all tasks; and working memory load savings or no change for all tasks.