Assessment of Mental Workload Using Physiological Measures with Random Forests in Maritime Teamwork

Abstract

Assessment of mental workload plays an important role in adaptive systems to perform dynamic task allocations for teamwork onboard. In our study, workload assessment models were established based on EEG, Eye movement, ECG, and performance data, respectively. The data were collected from team subjects operating maritime target identification and coping device allocation tasks collaboratively in a computer simulation program. Physiological measures were collected from wearable sensors, and the team workload was self-assessed using the Team Workload Questionnaire (TWLQ). Mental workload models were trained by the random forests algorithm to predict team workload with self-reported TWLQ measure as reference and physiological measures and objective performance measures as inputs. The low levels of MAPE (Mean Absolute Percent Error) suggested that these measures can be used to provide accurate assessment of operator mental workload in the tested type of maritime teamwork. This study demonstrates the possibility to assess operator status according to physiological measures, which could be employed in adaptive systems.