Continuous Assessment of Mental Workload During Complex Human-Machine Interaction: Inferring Cognitive State from...

Area of Science:

• Human-Computer Interaction
• Cognitive Science
• Aerospace Engineering

Background:

• Complex human-machine interfaces (HMIs) are increasingly prevalent in industrial and personal settings.
• Monitoring mental workload (MWL) is crucial to prevent cognitive overload, fatigue, and accidents in HMI operators.
• Professional helicopter pilots face significant cognitive demands during flight operations.

Purpose of the Study:

• To propose and validate a data-driven approach for continuous estimation of MWL in helicopter pilots.
• To compare the predictive power of physiological versus operational parameters for MWL estimation.
• To explore the development of intelligent systems for real-time MWL monitoring.

Main Methods:

• Development of a novel machine learning model trained on physiological and operational data.
• Utilizing realistic simulated flight scenarios for data collection.
• Evaluating model performance using metrics such as ROC AUC, F1 score, and PR AUC.

Main Results:

• The machine learning model achieved strong performance in MWL estimation (ROC AUC 0.836, F1 0.842, PR AUC 0.820).
• Operational parameters demonstrated superior predictive power for MWL compared to physiological signals.
• The study highlights the effectiveness of operational data in assessing pilot cognitive load.

Conclusions:

• The proposed data-driven approach enables continuous and accurate MWL estimation in HMI operators.
• Operational data is more relevant than physiological metrics for predicting MWL in this context.
• Findings support the development of intelligent systems for real-time cognitive load monitoring in aviation and other HMI-intensive fields.