Please use this identifier to cite or link to this item:
https://hdl.handle.net/10356/182335
Title: | Expertise level identification of air traffic controllers through visual measures and explainable AI | Authors: | Lin, Yen-Po | Keywords: | Computer and Information Science Engineering |
Issue Date: | 2024 | Publisher: | Nanyang Technological University | Source: | Lin, Y. (2024). Expertise level identification of air traffic controllers through visual measures and explainable AI. Master's thesis, Nanyang Technological University, Singapore. https://hdl.handle.net/10356/182335 | Abstract: | This dissertation explores the integration of eye-tracking metrics, machine learning (ML), and explainable artificial intelligence (XAI) tools to improve air traffic management (ATM) systems. Analyzing eye-tracking metrics like fixation count and duration of air traffic controllers (ATCos), the study could highlight significant correlations with expertise levels. Novices show high fixation counts and longer durations, which reflect scattered attention and cognitive overload. Experts, on the other hand, scan efficiently and demonstrate higher levels of situational awareness. Intermediate controllers exhibit transitional features that indicate developing expertise. Transparency in ML model decisions is ensured for actionable insights through real-time monitoring and training programs by the integration of SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME). These tools underline key features influencing the predictions that bridge complex models to operational applications. These results emphasize the added value of integrating physiological data in assessing the performance and cognitive load of ATCos. The key contributions are the development in using eye-tracking as a quantitative tool to assess expertise and the integration of XAI in transparent decision-making in ATM. Recommendations for practical implementation cover real-time monitoring systems for the detection of cognitive strain and adaptive training modules according to individual needs. Future research proposes the integration of additional physiological data and the use of advanced deep learning models to reach a granular analysis of behavioral patterns. This research showcases the potential transformative impact of AI-driven frameworks within ATM. Therein, it is asserted that AI increases safety, operational efficiency, and enhances training methodologies. This study lays the foundation for further innovations in integrating AI and human factors in safety-critical domains. | URI: | https://hdl.handle.net/10356/182335 | Schools: | School of Mechanical and Aerospace Engineering | Fulltext Permission: | restricted | Fulltext Availability: | With Fulltext |
Appears in Collections: | MAE Theses |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
Lin Yen-Po Dissertation.pdf Restricted Access | 2.38 MB | Adobe PDF | View/Open |
Page view(s)
45
Updated on Feb 11, 2025
Download(s)
1
Updated on Feb 11, 2025
Google ScholarTM
Check
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.