Fatigue-related situation awareness recognition using advanced machine learning

• Main Author: Li, Ruilin
• Other Authors: Wang Lipo
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Engineering
• Online Access: https://hdl.handle.net/10356/175644
• Institution: Nanyang Technological University

Situation awareness (SA) is a complex state of mind that results in physical and psychological changes on operators in various dynamic systems. Therefore, a precise measurement method of SA is the first priority for keeping the safety of the systems. Compared with traditional SA assessment techniques, Electroencephalography (EEG) has the key advantages of high time resolution, non-intrusiveness and being objective, which are more suitable for practical applications. This research aims to improve the end-to-end cross-subject SA recognition performance from the following aspects: dealing with the problems of subject variability, improving the efficiency of the backbone models, high complexity of input signals and data scarcity. Furthermore, multi-modal (EEG + eye tracking) SA recognition is also performed. Compared to the traditional calibration-based EEG processing, the methods for calibration-free and automated decoding from raw EEG signals are provided. This research is meaningful to the future practical use of EEG signals for SA recognition in various dynamic systems.