Temporal Relation Modeling and Multimodal Adversarial Alignment Network for Pilot Workload Evaluation.

Area of Science:

• Aerospace Engineering
• Neuroscience
• Biomedical Engineering

Background:

• Pilot workload evaluation is critical for flight safety.
• Physiological signals offer objective measures of mental states.
• Existing methods struggle with temporal dynamics and multimodal data fusion.

Purpose of the Study:

• To develop an advanced method for pilot workload evaluation.
• To address limitations in temporal modeling and multimodal data fusion.

Main Methods:

• Proposed a temporal relation modeling and multimodal adversarial alignment network (TRM-MAAN).
• Utilized a Transformer-based module for temporal relationship modeling.
• Employed an adversarial alignment module for multi-modal fusion.

Main Results:

• TRM-MAAN significantly outperformed baseline models in classifying pilot workload states.
• Achieved high classification accuracy and F1 scores across subjects.
• Demonstrated robust performance in integrating EEG and EMG data.

Conclusions:

• The TRM-MAAN offers improved accuracy and robustness for pilot workload evaluation.
• This approach enhances flight safety and has broad application prospects.
• Potential applications include clinical monitoring of fatigue and cognitive status.