Simulator-based evaluation of human response in emergencies

Abstract

In this context, this thesis aims to investigate human response in emergencies using an eye-tracker. To achieve this, the most common human-related tasks that lead to collisions in congested water were identified, and these tasks were employed as variables in an experiment conducted on a full-mission bridge simulator. This thesis presents a conceptual framework that integrates evidential reasoning (ER) and the standardized plant analysis risk‐human reliability analysis (SPAR-H) method to quantify human errors in practical studies. Fault tree analysis (FTA) is used to predict risk. The collision risk of ships in congested waters is selected as a case study to demonstrate the proposed method and show how a detailed analysis of collision risk can be obtained. The findings of the paper indicate that "inadequate watchkeeping due to sole lookout," "improper RADAR monitoring," and "ineffective execution of COLREG-related actions" are the most significant human errors contributing to ships' collision risk in congested water. The occurrence probability of the collision risk in congested waters was found to be 3.75E-01, which can be considered high risk. However, with sufficient lookout, efficient RADAR monitoring, and proper execution of COLREG actions, the occurrence probability of the collision risk in congested waters can be reduced to 2.55E-01. An experimental study was then planned to objectively explore the effect of the lookout on collision risk in congested water. Two sets of experiments were conducted to evaluate the cognitive process in a full-mission bridge simulator and determine seafarers' cognitive abilities and/or limitations. In the first experimental study, the determined accident scenario was adopted in the full-mission bridge simulator, and participants were tasked with passing through the Dover Strait under the condition of being the "sole lookout." In the second experimental study, a lookout with defined duties was added to the same accident scenario. The objective of the experiments was to recognize the influence of the lookout on collision risk in congested water under the bridge simulator environment. Additionally, the study explores human response during an emergency situation using objective (eye tracker) and subjective methods (questionnaires). In both scenarios, different participants were involved, and the effect of the lookout's presence on the bridge (a trained lookout with well-defined tasks in the scenario) on the participants' workload, visual attention, and collision risks when encountering multiple targets in congested water was examined. With increased bridge manning, lookout intervention reduced collision probability by 61.5%. The results show that lookout intervention helps reduce collision risk with the increase in bridge manning. Raw data from the Tobii Glasses 3 (eye tracker) were processed using the Tobii Pro Lab software. The software offers several outputs, from which five eye-tracking measures were chosen: Pupil Size Variation, Blink rate, Dwell Time, Fixation rate, and Saccades rate/89+6. After determining the eye tracker metrics related to workload and visual attention from the literature, the results were statistically analyzed. Although some metrics did not show a statistically significant relationship, the explanations of the results support the study's findings. Furthermore, a significant relationship was observed between the number of hits (number of fixations) features that were normalized in the first and second experiments and the state of the AOIs (Areas of Interest) determined during the design phase of the experiment. These AOIs include the Display unit, Top display unit, and Visualizations. The participants' visual attention was compared, and according to the statistical analyses, there is a significant relationship between the total number of fixations, fixation frequency, and visual attention (Display unit: p < 0.001, top display unit: p = 0.05, visualizations: p < 0.001). The Mann-Whitney U test performed on the two distributions, i.e., the first set of experiments and the second experiment, did not reveal any statistical differences related to average pupil dilation between the two groups (p = 0.3) regarding mental workload decrease between the groups. The Mann-Whitney U test results show no significant differences between the two groups' NASA-TLX scores of 6 dimensions (mental demand p = 0.14). The design stages of experimental studies are essential in obtaining accurate results, and this study describes a detailed experimental study process in the maritime sector. The experimental design processes and metrics used in this study provide guidance for further experimental research related to other ship operations using eye trackers. Real-time eye tracking measurement of seafarers' eye movement would be useful, providing feedback to reduce human errors. Eye tracking technology, as one of the objective measurements, is a relatively new approach in the maritime sector. The results of the study demonstrate that an eye tracker is a valuable tool for exploring human response in the maritime context. This thesis takes an in-depth initial step towards assessing human response in emergencies (specifically, a collision case study) using objective (biosensor) data with an eye tracker and subjective measures. The study is expected to make original contributions to maritime stakeholders, such as governmental bodies, classification societies, ship operators, and maritime researchers, in terms of understanding human response during emergency situations. Moreover, the proposed approach bridges the gap between objective and subjective data accuracy, thereby enhancing safety onboard.