An experimental study into the fault recognition of onboard systems by navigational officers

Autonomy has allowed the maritime industry to design integrated systems leading to the concept of maritime autonomous surface ships. As research towards autonomous operations increases seafarers must be equipped with the knowledge of how to react to onboard system faults and threats to the safety of crew, vessel and cargo. Consequently, the maritime industry may utilise bridge simulators to train seafarers in autonomous operations. By integrating simulation into navigational officer training, it is possible to aid the development of seafarers fault recognition patterns. Moreover, simulation training can provide seafarers with the knowledge to be proactive in fault finding over reactive. Therefore, this study is conducted in a navigational simulator and investigates the fault recognition patterns of seafarers during realistic watch conditions with alternative tasks i.e. paperwork. Moreover, a novel Event Tree Analysis method is proposed to analyse the performance of seafarers and effectiveness of human machine relationship. The study found a low percentage of candidates successfully reacted to all faults and without additional alarms the vessel may have resulted in further danger. Applying the methodology and data assimilated from the study could aid the development of navigational officer short courses, developing seafarers behavioural skills which complement their technical talents.


Introduction
As the maritime industry develops alongside current technological advancements autonomous transportation is a step towards the future for the industry. The maritime industry's direction of travel toward autonomy has been made apparent through various projects being undertaken to standardise the implementation of such systems onboard vessels. The Maritime Unmanned Navigation through Intelligence in Network (MUNIN) project assessed the potential systems to be fitted to vessels for autonomous vessel operations (AVO) development of the concept for unmanned autonomous ships and the implementation of initial programmes (MUNIN 2017). The introduction of AVO may then result in an increase of paperwork requiring to be completed by navigational officers. However, in 2016 a study found that the majority of seafarers feel that there is an increasing demand for documentation and paperwork to be completed and stored onboard. Furthermore, the results from the study highlighted the notion that seafarers had indicated that the increasing volume of paperwork may interfere with operating the vessel safely and was placing them under a significantly greater amount of pressure due to time constraints (Sampson et al. 2016).
In the modern day training regime for seafaring officers, simulators are a tool that while used in training standards is not heavily relied upon to aid the development of the watchkeeping skill of seafarers. With studies identifying the variation of exercises that can be undertaken such as pilot passage planning (Rønningen and Øvergård 2017) and how to improve lookout activity throughout a navigational watch (Youn et al. 2018) it is possible to further integrate simulators into the modern training regime of navigational officers. Additionally, research has outlined how to utilise simulators to advance technical and non-technical navigational skills (Sellberg et al. 2018 With the direction of the maritime industry heading towards autonomy utilising simulators in the seafaring training regime could prove useful in the transition to autonomous navigation as it will allow individuals to hone their skills in a real-time environment without the hazards experienced at sea (Baldauf et al. 2019).
The following study has been conducted which analyses a group of 50 individuals all studying to advance their careers in the navigational officer sector of the maritime industry and their reliance on current onboard automation systems and potential changes in situational awareness.

Increasing the use of autonomy
In recent years levels of automation have increased within the transportation industry. The benefits of moving towards full automation and then remote AVO are vast as they provide a level of safety and cost benefits that outweigh the benefits of having human operators onboard but only if the system is operated correctly (Strauch 2017). Nevertheless, automation like everything can experience malfunctions or if operated incorrectly then the automated systems can produce a level of danger to the operator vessel and environment. Research undertaken has shown that implementing automated systems on a basic level can ultimately result in degradation in situational awareness increase in automation bias and automation complacency (Pazouki et al. 2018). This reliance on the system can be extended to qualified officers as accidents such as the grounding of the Priscilla can be attributed to automation complacency among other key factors such as fatigue boredom and not following or utilising aids or systems provided to support navigational officers during their bridge watch (MAIB 2019). Present maritime operations are primarily conducted under the control and influence of a human operator. This includes expectations of the operator being able to make critical decisions and conduct bridge watchkeeping alarm management whilst maintaining control of the vessel in its daily passage. However, with studies showing that 85% of seagoing accidents may be attributed to human factors (Cordon et al. 2015) this poses the question 'Do the maritime industry and shipping companies rely too heavily on navigational officers?' As a result, by increasing the control of autonomous systems this will ultimately significantly impact the safety and performance of the vessel.
Despite the benefits of introducing AVO the systems will only function to a level limited by the ability of the operator. Therefore, should the operator fail to identify an error or input data incorrectly a fault will occur.

Simulation in maritime education and training
By learning from the aviation sector, the marine sector has begun to understand that using simulators in a training environment can improve operational safety as the simulator allows the operator access to a training environment where they are in control of the system and with correct programming can be susceptible to a variety of different hazardous exercises. This allows the operator to then mitigate against the dangerous incident emerging and deliver a feel of realism and training that the operator can then use when they are on the job (Aragon and Hearst 2005). Research conducted in 2013 by Sellberg Lindmark and Lundin analysed the student and examiner relationship whilst the students were undertaking a simulator based competency test. The results of the study indicated that some students needed and received supervisor support in the form of assistance from the examiner resulting in the student becoming certified in their use of equipment by COLREGs standards despite only completing the exercise with the examiners' care (Sellberg et al. 2018).
With research highlighting the impending challenges that autonomous shipping will bring to both shipping and seafarers regarding education and training (Baldauf et al. 2018) it has become apparent that changes will be required to validate maritime education and training (MET). As autonomous shipping looks to be introduced to the MET syllabus it is possible for the industry to utilise research work carried out within other transportation sectors and learn from their successes. In 2021 a study was conducted analysing the knowledge of MET which determined that the changes to the training programme should focus on cognitive communicative and operational skills; all skills which can be trained and developed through the use of maritime simulators (Emad et al. 2021). Therefore, the focus of this study was to assess officers' fault recognition skills through the use of maritime simulators. This then poses the hypothesis of this paper that subjecting a human navigational operator to standard wheelhouse-based distractions such as routine paperwork will result in a disregard for alarms and hazardous situations. Furthermore, analysing how current junior officers and cadets conduct their navigational watches may give an insight as to how they prioritise their work on the bridge.

Materials and methods
In this paper, a group of 50 individuals from the navigational section of shipping crews operating a simulated vessel within the wheelhouse were analysed. The study monitored each candidate and their own experience within the bridge. Each exercise was carried out using the Kongsberg secondary bridge suites. Each suite implemented the Kongsberg Polaris simulator software which allowed candidates to control the simulator and Seaview R5 visual software which gave the candidates a visual feedback representation of their actions when controlling the simulated vessel. The layout of the simulator included three screens in front giving the candidates a 120°view of the forward of the vessel and a screen at the backside of the simulator suit allowing candidates to view the aft and wake of the vessel. Additionally, the suites were equipped with a steering control unit a workstation with fire alarm control and a control console that included systems that would be expected to be located within the wheelhouse i.e. electronic chart display information system (ECDIS) radar and telegraph. The set-up of the bridge in the simulator is representative of a simplified view of the systems that are to be expected on board a live vessel with the main difference being the lack of port and starboard bridge wings. Figure 1 shows the integrated bridge simulator set up.

Experimental framework
In preparation for each experiment, a different exercise was created using the simulator suite. In addition to the simulator quizzes and a work pack were issued to the subjects within each test station to imitate basic paperwork. Which was expected to be completed on the bridge during the navigational officers' watch time.
To maintain continuity throughout each test station the same vessel and operational parameters were used which were preprogrammed into the simulation software. For the experiment, the ship chosen was a bulk carrier travelling at 14 knots following a course heading of 000. The vessel's autopilot had been configured to sound an off-course limit alarm once the vessel had exceeded a cross track limit of 1 nautical mile off-course which can be altered at sea depending on sea state and weather. The vessel's particulars were as follows: L.O.A -215 m Beam -31.8 m Draught -11.5 m Deadweight -22,691 Tonnes L.P.P -162.9 m Max Power -9160 kW Three 20 min exercises were designed with unique faults that would occur within each exercise to ensure authenticity and immersion in the simulator. Candidates would have to undertake all three exercises to ensure they could be assessed on each testing station. However, the order in which the candidate completed the tests was arbitrary and would be based on the candidate's choice. Before the start of the first exercise all candidates were given a familiarisation briefing. The briefing detailed how to operate the system in terms of controlling the simulated vessel and communications with the instructor. Subsequently, candidates completed all three exercises. The exercises were created to ensure that all candidates experienced: a mechanical fault in the form of a rudder offset failure (ROF); a series of alarms in the form of routine fire alarm testing; and an automation fault in the form of an autopilot gyro drift failure (GDF). Each exercise was given a different fault traffic condition and time stamp within the corresponding test station. Each fault that was designed had been cross approved by various navigational officers' experiences. The routine fire alarm testing is conducted weekly by the engine crew onboard. The rudder offset can be caused by a plethora of issues that may cause a jam with the steering gear. However, the gyro drift may be caused by a faulty set of batteries powering the magnetron of the system causing the gyro to wander which then would result in the vessels autopilot following an incorrect plot line. Different time stamps were issued to each exercise to ensure that the candidate was aware of the change of station.  Beyond the variables, all three exercises were configured in a manner to provoke the candidate to respond to errors and faults which occurred in the simulation. Visual cues in the form of cloud patterns star positions and the wake were in view. Resultant alarms were activated to allow the candidate to inspect the fault further at their own discretion and communications were set up between each test station and the monitoring station to create a feeling of supervision for the candidate allowing them to call the captain or anyone else they deemed relevant for the experiment. As a further measure every candidate was monitored using CCTV and microphones located in each testing station thus allowing the instructor to record and monitor the candidate's action throughout the exercise.
All exercises ran for a total of 20 min thus allowing the candidate to operate the simulator for an hour in total. The candidates were issued with a work pack upon entry to the exercise. In each work pack, the candidates received the following items: three answer sheets to complete in their corresponding workstations a ship particulars work sheet which they could attempt to complete over the course of their time in the simulator suites and a logbook with three exercise pages for them to highlight any abnormalities in the exercises.
By monitoring and analysing their work packs and debriefing them every candidate was able to convey acknowledgment of any abnormalities if detected within each exercise. For each exercise, variables were required to display the corresponding traffic vessels in the simulations. Table 1 shows the parameters for the variables. Additionally, the data shown in Table 1 correlates to the display plots shown in Figure 2(I), (III) and (VI) for each exercise respectively.

Exercise 1-0000 hours rudder offset
Upon entering the test station for exercise 1 the candidate will be presented with a darkened wheelhouse as the time stamp reads midnight. Figure 2(I) shows what the candidate would be able to see on the radar.
As the candidate begins the exercise, they will know that the vessel will travel at a course of 000 as per the orders of the autopilot. At 11 min into the exercise the rudder of the vessel will begin to offset to an angle of 7.5°to starboard. To add to the ROF the turning indicator will begin to freeze at 11 min in order to assess whether the candidate could recognise the fault using their own judgement. This will in fact hamper any manual operation and encourage the candidate to believe that the vessel may not be turning as the indicator is not moving. However, at 18 min into the exercise the indicator heading will unfreeze and the correct turning angle will be displayed. From the start of the ROF the magnetic compass will begin to make a clicking sound indicating that the vessel is turning furthermore the radars of the vessel will begin to indicate that the vessel is turning as the fault is purely mechanical and not systemic. The final visual cue to indicate the turning of the vessel is the position of the stars. Should the candidate look out of the windows onto the simulated sky they will begin to see that the stars are moving indicating that the vessel is no longer keeping a 000 heading.
Should the candidate leave the simulation running without altering the course the auto pilot alarm will begin to sound at 14 min and 56 s into the exercise. This will be the final prompt for the candidate to alter the course and acknowledge the alteration of heading for the vessel. Should the candidate proceed to not alter the course or take control of the vessel by the 20 min time limit the candidate will be given a time score of 540 s thus indicating that the candidate failed to recognise the fault. The radar plot in Figure 2(II) shows what the plot would look like should the vessels control remain untouched throughout the exercise.

Exercise 2-0800 gyro drift
When entering the simulator suite, the candidate will be presented with the radar display shown in Figure 2(III). As can be seen in the radar display plots there are three vessels within the proximity of the simulator ship.
The candidate will enter the simulator suite to find that the vessel is travelling at a heading of 000 as per the orders of the autopilot. At 9 min into the exercise, the vessel will begin to experience a GDF. The vessel will begin to deviate from its course at a drift rate of 3 degrees per minute until the vessel reaches an off-course limit of 20 degrees.
As the vessel begins to experience the GDF the vessel's magnetic compass will begin to start clicking thus indicating to the candidate that the vessel is deviating from its original course. However, as this error has affected the vessel's gyros the heading display and radar readings will deliver an output that the vessel is on a course heading of 000. During this exercise, the candidate will have to look closely at the positions of the surrounding vessels and use the tracking function on the radar to help them assess the situation. As the bridge is fitted with a backup gyro for redundancy the candidate may changeover to the vessel's second gyro and from there, they can clearly see that there has been a course deviation.
Should the candidate leave the simulation running without altering the course the auto pilot off track alarm will begin to sound 15 min and 54 s into the exercise. The sounding of the off-course alarm will act as the final prompt for the candidate to assess and attempt to correct the error. Should the candidate proceed to not alter the course or take control of the vessel by the 20 min time limit the candidate will be given a time score of 660 s thus indicating that the candidate failed to recognise the fault. Figure 2(IV) and 3(V) shows the radar plots of gyros 1 and 2 where gyro 1 shows the error display whereas gyro 2 shows the true course of the vessel.

Exercise 3-1600 fire alarm
When entering the simulator suite, the candidate will be presented with the radar display as shown in Figure 2(VI). As can be seen in the radar display there are two vessels in the proximity of the simulator ship.
Upon entering the simulator, the candidate will find that the vessel is travelling at a heading of 000 as per the orders of the autopilot. During this exercise, the candidate will not experience any faults which will put the vessel at risk of harm. At 1 min and 30 s the fire alarm panel will sound a fire alarm in zone 1 of the vessel however upon calling the captain and engine room the candidate will be told  that there is routine fire alarm testing taking place which will be carried out during the course of this simulation. The candidate will then experience alarms sounding every 90 s in the exercise thus enhancing the sense of alertness.
Due to there being no deviation from the course the vessel moves as expected. This can be seen from Figure 2(VII) which displays the final radar plot at the end of the exercise.

Analysis and results
The data analysed was the time taken for the candidates to react to the fault for the ROF and the GDF exercises. The statistical analysis was conducted for the following demographics of candidates: Age Rank and Education level. Collating the data into these demographics allowed for further analysis. Table 2 shows the range of candidates in terms of the aforementioned demographics.
To measure the reaction times for each individual candidate the candidate was monitored using visual and audio CCTV which allowed the instructor to record when the candidate reacted to the fault of the test station. Each candidate was given a reaction time ranging from the start of the fault 0 s to the end of the exercise 540 and 660 s for the ROF and GDF exercises respectively.

Raw data
The graph displayed in Figure 3 shows every candidate's individual response time to both the GDF and the ROF exercises. In the graph, the times at which both exercises finish are highlighted along with the times at which the autopilot off-track alarm begins to sound 236 s after the introduction of the ROF and 414 s after the introduction of the GDF.
From Figure 3 it can be seen that over half of the candidates successfully reacted to the fault in the ROF exercises. The total number of successful candidates was anticipated to exceed this value as the candidates should have a heightened sense of alertness due to the exercise being conducted in darkness. However, with correct prompting i.e. autopilot off-track alarm only five candidates failed to react to the fault.
Conversely Figure 3 shows that 14 candidates overall responded to the GDF. However, seven candidates required the sound of the cross track alarm before they reacted to the course deviation.
From Figure 3 it is also possible to see the volume of paperwork completed by each candidate. As it can be seen candidates that completed less paperwork were more likely to correctly recognise both the gyro drift and the rudder offset faults; the candidates who had the fastest reaction times for the exercises had all completed 5% of paperwork or less.

Simulator event tree analysis
An event tree analysis (ETA) was conducted for each exercise enabling each possible outcome and its probability to be analysed. The use of an ETA allows a logic diagram to be designed to analyse the sequential events deriving from the initial fault. Each diagram highlights the frequency of occurrence of each individual event (Punnoose 2018). A standard event tree is shown in Figure 4.
Each ETA shows an accidental event which can be defined as a significant deviation from the expected situation resulting in an unwanted consequence, therefore, leaving the candidate with multiple outcomes depending on their decision making. Each exercise consisted of a variety of level barriers that assisted the candidate similar to what the candidate would expect when conducting a navigational watch. The combinations of results provided by the ETA provide an insight into the variety of failure modes located within each exercise.
For each individual exercise the following method was conducted to ensure that the ETA was constructed accurately and therefore highlighting the safety barriers and outcomes: • Define the initial event that may cause an undesirable outcome • Define the safety barriers installed to negate unwarranted consequences • Design the event tree using the safety barriers and outcomes in sequential order • Identify the frequency of occurrence of individual events and probabilities of each outcome

Exercise 1-0000 hours rudder offset
From the results shown in Table 2 it can be seen that for age and education level the groups which had performed the most successfully were the '26-29' and 'Diploma' categories respectively. When assessing the candidate rank it can be seen that all candidates with a minimum of 4 months of bridge navigational watch experience successfully reacted to the fault. Conversely, the younger age groups were not as successful in recognising the ROF this may be linked to the lower success rates among the less experienced rank and education level categories. Figure 5(a) shows the probability of failure for each event of the ROF exercise. It can be seen that the largest percentage of candidates failed to carry out a main engine slowdown. Should the candidate have conducted a main engine slow down then the resultant outcome would have been reduced and the course deviation would have been minimised thus resulting in a successful yet non-desirable outcome.
When looking at the qualified officers only it can be seen in Figure 5(b) that all of the officer candidates successfully overcame all the safety barriers preceding activating the manual steering. However, it can be seen that the qualified officers faced difficulties when performing a main engine slowdown to assess the fault.

Exercise 2-0800 gyro drift
As seen in Table 2 all categorial groupings struggled to successfully react to the GDF with the largest number of unsuccessful attempts belonging to the younger and less experienced groups. The groups which had the greatest success in reacting to the GDF for all demographics were the '26-29' 'Degree' and both qualified officer groups. This may indicate that the situational awareness of the candidates increased with both age and experience. Figure 6(a) shows the probability of failures for each event throughout the GDF exercise. As seen in the diagram as candidates progressed through the exercise the probability of choosing the correct pathway was significantly greater than choosing the incorrect choice. However, of the 50 candidates the number who performed the exercise by choosing the correct pathway which resulted in the safest pathway for both crew and vessel was six correct attempts.
As seen in Figure 6(b) all officer candidates performed successfully until reacting to the spatial parameters of the exercise. Conversely, the candidates began to falter in requesting steering gear tests and assessing the back-up gyro both safety barriers which would have aided the indication of a gyro drift.

Exercise 3-1600 fire alarm
As shown in Figure 7(a) the event which most candidates struggled with was turning the fire alarm panel to test mode. By putting the fire alarm panel into test mode, the candidate would have been able to complete their work and conduct their watch safely without the consistent distraction of having to silence alarms individually.
As seen in Figure 7(b) all officer candidates successfully reacted and requested further details on the fire alarms. Moreover, the majority of candidates successfully switched the fire alarm panel into fire test mode thus acknowledging subsequent alarms.

Discussion
By implementing ETA, it was possible to further analyse the detail of how each candidate performed during the exercises. As shown in Figure 3, 45 candidates had successfully reacted to the ROF whereas 14 candidates reacted to the GDF. However, due to the complexity of each exercise using ETA allowed for a greater understanding of how candidates performed when confronted with different safety barriers which they had to overcome. Additionally, the ETA in Figures  5-7 show that despite there being a variety of potential outcomes an understanding can be developed of which event path was the most common among the candidates.
With the maritime industry heading towards implementing autonomous systems onboard vessels further training for navigational officers is evidently required. However, the potential for improvement is not solely limited to the improvement of seafarer training. By adopting a system that utilises a simplistic user interface and functionality that allows the operator to read error messages and act accordingly the role of a navigational officer will be better equipped to deal with the challenges ahead. The learning curve for navigational officers will be steep however by utilising current onboard navigational aids and equipment the curve may not be as steep as anticipated.
It is understood that the individual that had recognised the GDF fastest had recently completed a short course allowing them more time in the simulator prior to testing which could have resulted in a higher sense of alertness and familiarity with the software. However, to test this theory would require a further study regarding the impact of MET and short courses.
The ETA for officers allowed for a greater understanding of the success in choosing the correct path to avert the vessel from danger. However, it can be seen from the GDF and ROF exercises that officers are still susceptible to making similar mistakes to the less experienced candidates.
Situational awareness levels differed greatly among candidates. Introducing regular short courses may prove to be the key in development of an individual's situational awareness. Short courses aiding situational awareness currently exist one of such being the navigational aids equipment and simulation training management course. However, this course is only necessary for senior officers resulting in junior officers and cadets not reaping the benefits nor developing their situational awareness for isolated navigational watches.
For the candidates that did not recognise faults there are a number of variables that may have factored into their failure. One of these is being issued the paperwork to complete during the exercises. Paperwork is a necessity onboard vessels and can be a distraction as it takes the candidates attention away from conducting a safe navigational watch. From this study it can be seen that there is a possible connection between completing paperwork and trusting the system to operate correctly. Furthermore, when on watch seafarers need to prioritise the safety of the vessel at the expense of completing paperwork.
The transition towards autonomous navigation is rapidly approa ching. Therefore, the maritime industry should analyse two key focal points that will allow the successful development of AVO. The first being the utilisation and improvement of autopilot system and the second being collision avoidance. While the focus of this study is on the fault recognition skills of seafarers when encountering a fault, the results have shown a potential misunderstanding of the capabilities of the autopilot. Moreover, utilising collision avoidance algorithms in conjunction with fault recognition could not only benefit the seafarer but could also increase safety at sea.
While at sea operators are expected to work for two 4 h watches within a 24 h period. As these exercises are 20 min, they do not offer a full perspective of how operators may react when presented with an issue over the course of a standard 4 h navigational watch. Maritime simulators while proving to be a beneficial training tool can be limited in their capacity due to the operator being in a simulated environment versus a real world scenario. Additionally had the candidate pool been larger it would have been possible to gain a greater understanding of the possibility of situational awareness being more apparent in different candidate demographics. Due to the highly technical operations of automated systems the main issue derived from this study is the candidates bias towards systems such as the vessels autopilot. As highlighted within Exercise 2, 14 candidates successfully reacted to the gyro drift fault however 7 required the sounding of the alarm. With candidates displaying a level of complacency and bias as shown in this study navigational officers should be educated in behavioural skills such as these to ensure their success as autonomous systems are introduced. Conversely, this study has identified that while candidates may have displayed bias and complacency towards the automated system 45 candidates correctly recognised the manual rudder offset that if left unattended the vessel would have remained off course.

Conclusion
Utilising bridge simulators as a training method will aid the skill development of navigational seafarers. Autonomous technology should be introduced in simulators prior to being adopted by vessels to allow seafarers to conduct simulated watches in real time without the dangers one can encounter at sea. Furthermore, this will permit navigators to become familiar with and be assessed on autonomous equipment prior to operating the system in a real world environment.
With 70% of test candidates failing to identify any faults within the GDF exercise it is possible that the candidates have demonstrated a reliance on the vessel's autopilot placing more trust in the system than in their own abilities and observations. This complacency or bias could have affected the candidates' watchkeeping routine i.e. assessing the magnetic compass.
From the exercises conducted it is apparent that older and more experienced candidates displayed less trust in the autopilot system and were more observant thus displaying a higher level of situational awareness. For all of the exercises, candidates were given multiple opportunities to recognise react and correct the fault however only a limited number of candidates successfully completed each exercise.
Adapting a current short course or introducing a new one would be a key step in the process towards increasing the levels of situational awareness displayed by young seafarers. Additionally introducing a situational awareness short course at an early stage of the navigational officer curriculum may develop situational awareness among junior officers leading towards implementation of AVO.
However, for navigational officers to become more adept with future systems it is crucial that they should be experts with the current onboard automated systems such as the autopilot. From the study conducted only 50% of qualified officers reacted to the automated system failures within the allotted time frame. As a result, this paper has identified a large variance in skill between officers of the same rank which may indicate that automation bias is prevalent among the candidates tested.
Utilising ETA it was shown from the GDF exercise that of the officer candidates only 34% successfully identified the fault and undertook the correct course of action to avert the failure. Therefore, it is apparent that although the qualified officers were more successful in fault recognition and emergency protocols they were not immune to displaying automation bias and a degradation of situational awareness. Additionally bridge watch navigational alarm systems which benefit the situational awareness of a navigational officer can be deactivated. The deactivation of such systems can result in hazardous consequences and accidents.
For this study a possible reason that contributed to the failure of fault recognition is the introduction of paperwork. This can be seen from both the GDF and ROF exercises where candidates failed to recognise the cause from the final off-course alarm assuming that the alarm was nothing to worry about. This proves the hypothesis of the paper that when subjected to distractions on the bridge such as other duties human operators are susceptible to making errors and disregarding bridge alarms. The focus for the maritime industry regarding autonomous navigation has been towards a consolidated bridge alarm system and the installation of navigational sensors to aid and alert the operators. Enhancing the bridge alarm system to give seafarers the opportunity to explore the cause of a fault while suitably alerting them could be beneficial. However, the argument could also be considered that there may be an issue of alarm fatigue which may dull the senses of seafarers over long periods of time. Incidents such as the Priscilla discussed in this study demonstrated that seafarers have sufficient warning systems in place to help correct potential incidents from occurring, but it is the duty of the individual to undertake the correct procedure to avert danger.
Full autonomous shipping will not be introduced overnight. Research has indicated that while there is an enthusiasm for the introduction of autonomous shipping it will be gradual allowing time for the maritime industry to adapt. It is imperative that navigational officers have suitable training with regard to management of autonomous systems onboard.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Notes on contributors
Jevon Philip Chan is a PhD candidate at Newcastle University in Newcastle Upon Tyne, UK. Following the completion of his marine engineering cadetship he furthered his knowledge within the maritime industry by completing his Bachelors degree in 2019, where he then progressed to becoming a PhD candidate. The background for his PhD research is assessing the behavioral skill development of navigational officers and safety aspects of autonomous shipping. Moreover, his research has lead him to analysing the impacts which autonomous shipping may have towards the merchant maritime sector and how seafarers will cope with the transition.
Kayvan Pazouki is a senior lecturer in Marine Engineering. He has been a seagoing marine engineer for nine years, before joining University. His research interests include development of performance monitoring tools through physical and/or inferential measurement systems, Human automation interaction and shipping environmental indexing systems.
Dr Rosemary A. Norman is a Senior Lecturer in Marine Electrical Systems at Newcastle University. Following her PhD, she spent 10 years in industry before joining the university in 2004. Her research interests include automation and underwater vehicles, marine renewable energy and marine applications of data analytics and machine learning.