Date of Award

12-2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Engineering and Sciences

First Advisor

A. Lucas Stephane

Second Advisor

Deborah S. Carstens

Third Advisor

Winston E. Scott

Fourth Advisor

Ralph D. Kimberlin

Abstract

General aviation (GA) loss of control accidents, in the United States, have continued to occur despite efforts by government and non-government research agencies. The lack of data during loss of control events (due to the absence of data recorders in small aircraft) does not allow for an accurate reconstruction of the accidents, causing accident investigators to focus on generic causes like pilot error. Flight test research has shown that there are various systems induced causal factors that can potentially set up an aircraft to depart controlled flight and do not aid pilot efforts. An example of these causal factors is the flap configuration change on certain GA aircraft. Although there is sufficient data on aircraft systems, there is a lack of data regarding pilot cognition in single-pilot operating airplanes, calling for an exploratory study to design a method to complement the already accumulated knowledge (through systems flight testing) with human-focused data. This study provides a novel flight test method that allows for human-systems integration between aircraft dynamics (the behavior of the aircraft), and pilot cognition (the behavior of the pilot). The dissertation contributes to the field of flight analysis and human-systems integration by exploring this new method that integrates the Critical Decision Method (CDM) cognitive modeling technique with virtual reality simulations. Given the critical and dangerous nature of loss of control, this study provides a high fidelity, high realism simulation environment in VR. The flight test method involves a novel framework that requires knowledge elicitation (KE) sessions in the form of protocol analysis and critical decision method, validated through integrated cognitive mapping during VR simulated flight tests. To consolidate the VR simulation experimental setup, this dissertation also includes usability tests to demonstrate the applicability and validity of VR as a substitute for traditional aerospace simulation methods. This study is divided into three main phases: cognitive modeling through KE, usability tests for VR use in aerospace research, and cognitive modeling through VR flight tests. The KE sessions included six expert pilots, the usability nine expert pilots, and the VR flight tests included eight expert pilots. The expert pilots were all certified flight instructors (CFIs) except for one GA domain expert. The cognitive modeling through KE provided fundamental data in understanding the cognitive functions present during traffic pattern operations. The cognitive modeling through VR flight tests (with aircraft dynamics) showed however a difference in the way the descent is performed. While the KE results focus more on “mental simulations”, the VR tests reported a dominance of “mental model development”, indicating that the traffic pattern is flown as a unified flight segment. The cognitive modeling provided in this document also describes niches for LOC, especially with details on flap configuration changes. It was found that trim change should be more system-centered and less human-centered, and that LOC can be mitigated through the use of energy management training in GA, through adequate aircraft design changes (i.e. modifying aerodynamic parameters by modifying the actual design), and through the use of situation awareness enhancing technology like augmented reality visuals. Finally, recommendations are provided to improve the method and collect further data to continue the efforts of mitigating LOC in aviation. The novel method was also identified as a means of enhancing (or defining) design requirements for multiple applications (e.g. aviation, automotive, control rooms, etc.), training (e.g. using expert cognitive modeling as a baseline for proper cognitive activity and identify lackings of student pilots’ cognitive activity), and even automation (including artificial intelligence - AI).

Comments

Copyright held by author

Share

COinS