EEG-based workload evaluation in a ship's bridge simulator based assessment

• Main Author: Low, Gabriel Min Keong
• Other Authors: Olga Sourina
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/68560
• Institution: Nanyang Technological University

This is a novel research project in collaboration with the Singapore Maritime Institute Simulation and Modelling (S&M) R&D Programme to study Human Factors in Maritime Risk Management. This project aims to study the workload performance of cadets using a ship simulator and together with an electroencephalogram (EEG) – a brain based state monitoring tool. This study was initiated as there are numerous human-factored-induced accidents in the maritime industry. Many studies and research to negate further accidents from occurring do not target the root cause of these accidents; the human element. With the use of EEG and real-life simulations and human interactions, reactions to a problem can be better studied and analysed to try reduce or prevent further maritime accidents from occurring. EEG detects electrical activity in the brain by using tiny, electrodes that are attached to the scalp of a subject. Communication amongst brain cells are through electrical impulses and it happens very actively, this is displayed by undulating lines on EEG recordings. EEG data have been collected from 7 subjects (subjects 5 to 11) with varying experience and age from the Singapore Maritime Academy (SMA) during simulation runs with 4 different difficulty levels. The data will then be processed using various algorithms in Matlab and Python programs. The data will be partitioned, features will then be extracted using Higuchi fractal dimensioning and statistical features and lastly these features will be trained and used as a classifier to determine the different workload levels present during the simulation. Once workload levels are determined after processing is done, workload levels are analysed in a few ways. Average workload values are determined for each subject in each exercise, to see if there is any positive correlation with workload levels and exercise difficulty. Video recording is analysed together with workload levels at each data point. The hypothesises of this report is 1) to determine if workload levels increases with the difficulty of exercise, 2) whether is there an increase in workload level when a scenario is present and 3) to determine if the highest recorded workload level is reflected when a major accident occurs. There are mixed correlations in relation with exercise difficulty and workload levels, subjects 5, 9 and 10 show positive correlation with workload level and exercise difficulty (workload level increases with increase in exercise difficulty) subjects 6 and 11 show negative correlation (workload decreases with increase in exercise difficulty). Subject 8 did not complete all exercises and Subject 7 did not have complete data. From this, it shows a mixed result with respect to hypothesis 1 (both positive and negative correlation). From the in-depth analysis of subjects 5 and 11, workload increases when a scenario is presented in the simulator and the highest level of workload can be seen in subject 11 when the subject got into an accident. Hypothesis 2 and 3 are fulfilled through this analysis. Recommendations for future exercises are also discussed after conducting the preliminary study. Some would include increasing the sample size of subjects, ensuring that there are no external factors that may interfere with the exercise.