A Systematic Review of Current Adaptive Human-Machine Interface Research
A comprehensive systematic review examining how intelligent interfaces can dynamically adjust to operator cognitive state, workload, and situational awareness in unmanned aircraft systems. Analyzed 108 peer-reviewed samples across three primary categories: workload management, situation awareness, and autonomy.
Ryan A. Blakeney
Embry-Riddle Aeronautical University
UNSY 691 Graduate Capstone Paper
May 2020
Abstract
Adaptive Human-Machine Interfaces (AHMIs) represent a convergence of sensor technology, machine learning, and human factors engineering. This systematic review analyzes 108 peer-reviewed journal samples to establish the predominant definition and standards used in the literature for adaptive interfaces in unmanned aircraft systems. The research categorizes existing work into three primary domains: workload management, situation awareness, and autonomy, with further sub-categories established through manual and computer-assisted classification.
Introduction
The literature review is focused on Workload Management, Situation Awareness, and Autonomy to match the three categories for the research. These three categories had the most prominent presence among the samples used for the study. A literature review focusing on these topics will show the diversity in research in Adaptive Interfaces.
Sensor-equipped control stations know about an operator's activities, preferences, and previous interactions that provide data that can be used by the control station to be proactive and anticipate the operator's actions, needs, and preferences. The sensors vary based on the type of control station and the type of operation. Examples of sensor technology are eye-gaze tracking, voice recognition, and an electroencephalogram (EEG). These systems can be utilized together or separately to monitor the operator and measure their mental state.
As the mental workload is measured, the operator display changes to manage the operator's workload. The complexity of the topics in this literature review contributes to the variability in terms, subjects, and definitions of Adaptive Interfaces. Workload, Situation Awareness, and the algorithms to support the two are all labeled in the literature review as an Adaptive Interface.
Research Method
Study Design
The study's design included an extensive search of related material associated with Adaptive Human-Machine Interfaces and a meta-analysis of the pre-existing data. Research related to the subject of adaptive interfaces was examined and categorized to determine the prominent definition and standard for adaptive interfaces.
Data Collection
Data for this research comprised journals and books related to the topic of "Adaptive Human-Machine Interfaces." The published information indicated that it was peer-reviewed or published by an accredited university. The published work included adaptive interfaces used in manned and unmanned systems.
Data were collected using the online journal resources ERAU Hunt Library Eagle Search, Sage Journals Database, ResearchGate, Science Direct, Springer Link, Google Scholar, and IEEE Xplore.
Categories and Sub-Categories
The journals were categorized into three categories of Workload, Situation Awareness, and Autonomy:
Workload Management: Measurement of workload, use of eye-gaze, use of voice, and use of EEG.
Situation Awareness: Measurement of operator SA, loss of SA, and Situation to change HMI.
Autonomy: AI, automation, and machine learning.
A Chi-square goodness of fit test measured the 108 journals' categories to indicate if the journals are equally indicating the same definition and standards or if the research favors one topic over the others.
Summary
The literature review provides information on how Adaptive Human-Machine Interfaces can gauge an operator's SA while modifying the display. The changes in how much SA the operator has during a mission or flight is measured using technology such as voice or EEG. Adaptive Human-Machine Interfaces can also measure operator workload to determine how much SA the display should provide at any given time.
To determine how and when to change the display, automation is used to determine how much to change, when to change the display, and what to show at any given time using machine learning or artificial intelligence.
Conclusion
This systematic review establishes a framework for understanding how adaptive interfaces are defined and implemented in unmanned systems research. The three-category structure (Workload Management, Situation Awareness, Autonomy) provides a taxonomy that can guide future research and development in adaptive human-machine interfaces for UAS operations.
Limitations and Delimitations
A limitation of this study is the specificity of the topic and the interpretation of the categories and sub-categories. Research into adaptive interfaces is new and is still ongoing. The qualitative nature of the study means that not all material associated with this topic was found due to limitations in access to all available information.
A delimitation of this study is the focus on unmanned systems. The research will focus on unmanned systems but will use published work not directly related to unmanned systems to ensure the research project does not ignore relevant work related to the project.
Originally submitted to Embry-Riddle Aeronautical University as part of the MSUS program.