See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318655874

Naval Artificial Intelligence

Conference Paper · June 2017

DOI: 10.1201/9781315099132-51

CITATIONS READS

12 3,472

2 authors:

Eric Kulbiej Piotr Wołejsza

10 PUBLICATIONS   29 CITATIONS   
Maritime University of Szczecin
27 PUBLICATIONS   134 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

AVAL - Autonomous Vessel with an Air Look View project

NAVDEC View project

All content following this page was uploaded by Piotr Wołejsza on 26 June 2018.

The user has requested enhancement of the downloaded file.

Naval Artificial Intelligence
E. Kulbiej & P. Wołejsza
Maritime University of Szczecin, Szczecin, Poland

ABSTRACT: The following paper focuses on an analysis of a comprehensive system of nautical vessel’s
artificial intelligence. Such a complex structure should compromise a number of parallel major and minor
items of vessel’s equipment in order to maintain correct management of the vessel itself. Stemming from
individual interests of the authors, an attempt to generalize a theoretical basis for a naval artificial intelligence
(NAI) is presented. The definition of NAI and its components is derived and discussed. Schematics of an
algorithm is proposed; the NAI as a virtual managing party of a vessel’s deck and engine department,
maintaining safe navigation and the freightage of the cargo as well as harbour operations.

1 INTRODUCTION System (NDSS), an example of which is Navdec,

system created under the research team of professor
Astral navigation was, for many years, one of the Pietrzykowski from Maritime University of Szczecin
main methods of determining ship’s position. The (Wołejsza P., 2013; Wołejsza P., 2014).
twentieth century brought a lot of invention on board The future upholds an increased need for
the ships e.g. radar. Due to process of automation, technological advance, so that congestion of
the size of a vessel’s crew has been continuously merchant transport is satisfied. In order to achieve
reduced, nowadays oscillating around the number of that goal, brand new technologies might be
15 crewmembers, for an average merchant ship. necessary, such as an autonomous vessel. Such a
More and more systems providing assistance for the unmanned ship is the ultimate stage of technological
navigational bridge team are being developed and development of maritime merchant branch of
implemented, for instance Automatic Identification economy. This paper aims to propose a system
System (AIS), Automatic Radar Plotting Aid called Naval Artificial Intelligence (NAI), which
(ARPA) or even Electronic Navigation Charts should act as a virtual operator of a ship, providing
system (ENC). Currently the youngest and most full monitoring and conducting all sorts of services,
novel invention is the Navigational Decision Support commencing with ensuring safety of the object.

Figure 1. Master’s duties onboard a ship (own study).

Figure 2. Naval artificial intelligence definition in form of a graph (own study).

2 SYSTEM OF NAVAL ARTIFICIAL recorded and reported to the ship-owner and their
INTELLIGENCE remarks are to be taken into account in the
navigational planning.
The artificial intelligence is described as the Own ship maintenance sector of the system ought
capability of a computer-driven machine to imitate to oversee ship’s hardware, that is to say, provide
intelligent human behaviour. Its main objective correct use and ensure non-deteriorating
would be to ensure navigation of the vessel, which is performance of the main and auxiliary engines of the
to be described as a safe voyage between the point of vessel. System of comprehensive detector would
departure and the point of arrival whiles providing map the whole vehicle creating network of probes
transfer of untouched cargo. This duty, main in the that sufficiently supervises hull, mooring equipment,
means of modern seaborne transport industry, is electrical installation, ballast and other tanks’
currently undertaken on average by a crew of fifteen installations, fire-fighting installations and thus
people. Divided into two main departments: deck prevent physical maladies, which could render the
and engine, they control and maintain the vessel. The ship inoperative due to one of aforementioned
ship’s master is the person directly entitled by the installation compromise.
ship-owner to fulfil mentioned goal. He holds the
immediate responsibility to provide several
distinctive functions, as far as vessel’s exploitation
is concerned. They are shown in the figure 1.
It is clear that the master is the overseer of the
vessel. Thus, module of artificial intelligence
employed to run the ship should imitate an
experienced captain’s behaviour as it would directly
inherit his major duties. But then again, master’s
duties limits to supervising and delegating most of
the tasks, especially when it comes to physical Figure 3. Item-tree of NAI (own study).
activities. This phenomenon is not rare and rather
typical of employees on the management level.
In order to achieve the premise of safe and
effective voyage, vessel controlling system should
execute more complex variety of tasks. It can be
stated that Naval Artificial Intelligence is a system
that undertakes navigational tasks, provides
exemplar ship’s maintenance, executes safe cargo
Figure 4. Physical model interpretation of NAI (own study).
handling, transportation and shore operations and
establishes successful security of ship, while
preventing from rendering herself dangerous to For the cargo transport using seaborne methods is
environment and other objects. This definition by a the key feature ships are used it is vital that NAI
mean of a graph can be visually described as in employs sufficient means of cargo protection and its
figure 2. loading/discharging. In these means mooring system
Vessel’s navigation sector will include mainly is required and the holds need to be operational in a
tasks connected to widely-known marine navigation way that meets the harbours equipment’s
but is not to be limited by it. Therefore NAI is to requirements. For the purpose of analysis of NAI it
declare the most effective route from the point of is to be stated, that should proficient mooring system
departure to the point of arrival. The route itself is to be unavailable, NAI may use external help of shore
be under constant scrutiny of the system’s facilities and workers. Then role of ship during the
optimisation algorithms so it can be altered on a cargo transferring process can be either maintained
regular basis, dynamically answering to situation. by NAI or remotely dispatched to shore based office.
The voyage should be moreover continuously The last but not the least group of NAI functions
is security segment. In contrast to safety of cargo, in
regards to safety and security of the ship NAI should crucial. Provided that the NAI system declares
provide that route undertaken by the vessel is free of evolutionary example of intelligence module it may
threatening encounters. In other words NAI has to, execute a risk-minimalizing method based on
by successfully implementing Rules of the Road dynamic and logical approach to the case. Important
(COLREG), omit unnecessary closest points of feature of NAI seems to be self-learning module and
approaches, undertake last moment manoeuvre is
connection to a broader network of similar systems.
necessary and steer the vessel in respect to other
watercrafts. In special situations NAI driven vessel Thanks to that, once encountered, the problem is
should deploy itself for rescue mission and proceed then solved with automatic response should it
to assistance for any object or person in need. The reappear. The block schema of this problem solving
actions executed by NAI ought to be deprived of algorithm is described in figure 5.
losses, but if these unfortunately are to occur, the
damages to cargo, environment etc. are to be
minimalised by all means. The situation when on-
board a physical malady appears, for instance a fire
or a leak, NAI should be provided with installed
equipment or other means sufficient to cope with
encountered problem.
Aforementioned particular and exemplar pieces
of equipment and facilities that comprises for the
whole item-tree of NAI are graphically presented in
figure 3. On the other hand, physical model
interpretation of NAI is shown in figure 4.

3 BEHAVIOUR MODEL OF NAI Figure 5. Problem solving algorithm (own study).

In contrast to typical navigational equipment that is Provided no more mission threatening problems
already available for maritime officers, NAI not only are encountered, NAI is to proceed to decision
downloads data from different sources such as GPS making component of own artificial intelligence
and displays them after calculation but also applies o module. In this step the situation need in-depth
wn means of understanding. It is to say, that the data investigation in search for several particular (mainly
is processed and checked so all effects that may have numerical) traits. When the investigation is
a negative impact, for example relativistic effects complete, the mission optimization process is
(Kulbiej E., 2016), are taken into account. But then commenced. The purpose of this action is purely of
again, the facilities and gear installed on the vessel economical basis, as everything undertaken by NAI
are only limbs, futile and otiose should be should lead to minimalizing costs of the voyage and
abandoned. It is NAI that functions as a brain of the cargo handling. For instance, the route is maximally
vessel with presumably main and auxiliary engines shortened after taking into account untypical needs
in the role of heart and optical fibres or other ways of weather navigation. A fully planned route with
of convening data as veins. To serve its purpose of pre-set waypoints and an initial report is regarded as
problem solving, decision making and way leading a fruit of this activity. The report is to be presented
NAI must be given within the ship a sort of to the on-shore human crew, however the act of
authority, not to mention sufficient means or reporting should be executed by NAI on regular and
autonomy. continuous basis. One ought to never underestimate
As a system of artificial intelligence, NAI is to the fact that NAI is only artificial machine operating
solve encountered problems on the premise of system.
several means of solution acquisition. Mainly and The way leading activity of NAI is the continuous
firstly the problem is to be compared to expertise control over the vessel, namely deployment and use
of provided equipment on-board accordingly to
based knowledge bank. If a suitable solution for the
encountered problem or simply to execute tasks
problem is found, it can be directly applied and the decided on in the decision making section. That
threat is compromised. Otherwise, one of means cooperation between different parts of
preinstalled algorithms modules use is proposed. machinery in order to fulfil the goal of cargo
This is the autonomous step of decision making in transportation.
contrast to previous automatic approach. If the The whole behaviour tree of NAI is proposed in a
problem is still unsolved, further means are figure 6.
applicable. Specially designated shore personnel
might be contacted, however immediate response is
 of collisions involving ships caused by human error
is very high and is equal from 75 to 96% depends on
time and a region of the world (Antão and Guedes
Soares, 2008; Celik and Cebi, 2009; Harrald et al.,
1998; Zhang et al., 2013a, 2013b). Algorithm does
not become nervous in collision situation. It does not
feel stress. It will generate anti-collision manoeuvre,
which will be executed by autopilot and collision
will be avoided. There many examples, when
collision took place because of nervous action of
navigators (Kulbiej E., Wołejsza P., 2016). In cited
example, vessels would pass each other without any
manoeuvre. Unfortunately, due to “coordinated”
action of both navigators, an accident took place
entailing many casualties and loos of m/v Baltic
Ace.
Figure 6. NAI behavior tree (own study). Second of all reduces or eliminates salaries of
crewmembers. One can say, that it can bring
unemployment among seamen, who are over one
4 COMPONENTS OF NAI million throughout the world (IMO, 2016)
Around half of them are officers both from deck
There are following components proposed in the or engine department. Estimates of the IMO
NAI: (International Maritime Organization) and ILO
(International Labour Organization) show that in
2020, will be missing about 40,000 officers of the
4.1 Autonomous vessel, which consists of three merchant fleet. This is due on the one hand less
main elements: unmanned navigational bridge, popular profession (UK Government, 2011)
unmanned engine room and communication from the other with an increasing number of
protocol. vessels that transport already about 90% of all goods
Nowadays most of existing ships have in world trade. Lack of 40,000 officers, means that
“autonomous” engine with A24 class, which means 5,000 ships have no staff. Those can be the pilot
that all processes can be executed without crew. group of autonomous ships.
Alarm panels distributed throughout the ship inform Third, but not least, algorithms can reduce fuel
engineers of any abnormal activities within engine consumption. Medium size container ship
room department. Similar panel can be installed in (8,000TEU - Twenty feet Equivalent Unit) consumes
remote control centre located ashore. If any problem about 260MT HFO (Heavy Fuel Oil) per day at full
arises, it can be solved using procedure presented on speed, which means that it burns 180l of fuel per
Fig.5. minute. Thanks to the optimal anti-collision
Navigational bridges are still required to be trajectory, shorten the time to complete this
manned at least by an officer. Usually at night watch manoeuvre. In practice, the navigator assesses the
is doubled. Same situation is in restricted visibility, size of turn basing on own experience. When the
which from the point of view of common sense has assessment is incorrect, improves own manoeuvre,
no practical explanation. Visual observation carried but each rudder inclination is a measurable loss of
out by watchman will not improve safety of ship. In fuel. Assuming that collision-avoidance manoeuvres
such situation the officer relies on information is shorten only a minute a day, we can save 65 tons
received from radar, AIS and other electronic of fuel per year, which at a cost of about USD 320
equipment (Pietrzykowski Z., et al., 2013). All these per MT, gives savings of over 20.000USD, but in the
information can be transmitted to remote control case of low sulphur fuel savings are almost the
centre located ashore. The decision how to continue double.
voyage or how to avoid collision can be taken using Above mentioned advantages will definitely
procedure presented on Fig.4. If navigators have reduce cost of sea transport. As most of the cargo are
taken decisions basing on electronic sources in transported using this mode, total cost of goods will
restricted visibility, they can also rely on these be reduced. To execute such a goal, an effective
sources in good visibility. This is why, in authors communication protocol should be implemented.
opinion, such decisions can be taken by NAI, which Transmission between ship and shore can be
is supervised remotely from control centre located executed via satellite. Speed which is presently
ashore. Proposed solution has a lot advantages. First available, enables video streaming and in coming
of all we can eliminate human factor, which is the years will be even faster.
main reason of accidents at sea. For example number
 In authors opinion, satellite communication will Such information should be transferred to AIS an
not be so effective for ship to ship information transmitted by AIS automatically, because the
exchange. First of all it is not free of charge. It has navigator in collision situation has no time to do so.
relatively long delay, which can be important This information is crucial for another vessel/vessels
particularly during last moment manoeuvre. involved in collision situation.
Moreover there are existing systems, which after Solution: direct and two way transmission
some improvement, should meet requirements for between anticollsion system and AIS
autonomous vessels (Pietrzykowski Z., et al., 2014). The authors find also great potential to create
Among them AIS and DSC (Digital Selective local communication platform in NANET network
Calling) are considered. Both systems enable to send (Nautical Ad-hoc Network), which can use vacated
short ship to ship messages. by analog TV radio frequencies. NANET uses Wi-Fi
protocol to create ad hoc network. Due to 700 MHz
Table 1. Available slots in AIS standard (IALA, 2009, 2011)
__________________________________________________ frequencies, network has a range of at least 20 Nm,
Message Spare bits Title
__________________________________________________ which is more than enough from the anticollision
1,2,3 3 Position report perspective.
4 9 Base station report Integral parts of the system being developed for
5 1 Ship static and voyage related data automatic communication will be the language of
9 7 Standard SAR aircraft position report communication, protocol and protocol interaction
14 968 Safety related broadcast message and negotiation strategies (Pietrzykowski Z., et al.,
18 8 Standard class B equipment position
report 2016).
19 8 Extended class B equipment position
report
21 1 Aids to navigation report 4.2 Berthing module
__________________________________________________
From the technical point of view, autonomous
They can be used to convey intentions or agree passage of the vessel from point A to B is relatively
manoeuver. Crucial issue is that such messages have easy. Challenge arises, when other objects, which
to be sent autonomously. For example when were not predicted in the sailing plan, obstruct ship’s
anticollision system generates manoeuver, message movement. Next challenge arises, when ship is
with intended action have to be transmitted to other approaching destination port. Basing on current
target within selected range e.g. 10 Nm. In other technology, it is hard to imagine, that ship can berth
case, when avoiding collision is not possible by without crew. Additionally, it is required by local
manoeuvre by one ship only, the coordinated authorities, that most of the ships have to be boarded
manoeuvre has to be transmitted to dangerous target. by pilot, before coming alongside. Combining above
It has to be, without delay, executed by mentioned facts, temporally solution for transitional
rudder/autopilot. period, can be introduced as follows: mooring team
What can be added to existing standard: can board the ship together with the pilot on the
 Route to the next port of call, including all roads and conduct the ship from the roads to the
waypoints (WPs) or at least 5 next WPs – destination berth. Team for unberthing will board the
transmitted by each ship in Message 5: Ship ship alongside and sign off together with the pilot to
static and voyage related data. Unfortunately there pilot boat or helicopter.
is only one spare bit in msg 5 in two slots. When the ship is already alongside, the process of
Solution: to add third, fourth, fifth etc. slots to discharging/loading the cargo can be started.
this msg to accommodate a/m information which are
crucial to predict behaviour of other ships 4.3 Cargo module
particularly in collision situation.
 Meteorological and navigational warning – There are already many terminals, where the cargo is
transmitted by each ship in Message 14: Safety handled by autonomous machines. Such container
Related broadcast message. There are 968 spare terminals exist in Hamburg or Rotterdam (Delta
bits in msg 14. It gives 161 available characters Terminal, 2017).
(6-bits ASCII) to send a/m information. Also wet products are handled with limited
Unfortunately this option is very seldom used by presence of human. After connecting to manifolds,
vessels to transmit a/m information. It’s usually all the process are controlled by computers with man
used to transmit test msg or “social” information. supervision. It is also easy to imagine, that cargoes
Solution: awareness of seafarers about the in bulk can be transferred from/to shore by
opportunities. autonomous cranes and conveyors.
 Anticollision manoeuvre – transmitted by each Challenges arise during sea passage. Some
ship in Message 14: Safety Related broadcast cargoes require monitoring (temperature, humidity
message etc.) and special care e.g. ventilation. We can image
that such process is controlled by NAI, but what far behind, either on the side-lines of the
happens in case of motor failure in reefer container? mainstream. The main limitation is the international
regulations, and generally tedious and long-lasting
process of changing these provisions or to introduce
5 VESSEL’S MANAGEMENT new regulations. However, the steady increase in
tonnage while reducing the number of people willing
Low interest in the seafaring profession, shortages of to work as a seaman, accelerate the process of
highly qualified officers and still a high proportion change that from a social point of view, are highly
of human error in the causes of marine accidents. All desirable. The sailors will work on the land. Long
these factors result that shipowners think about lasting contracts away from home. This in turn
replacing human by machine. There is currently no should reduce the cost of transport, although the
legal possibility to implement it. Shipowners shall authors expect that in the initial period, the cost will
continue to apply stringent Safe manning be higher, as even on the mentioned Delta terminal,
certificates, which precisely regulate the where the introduction of autonomous systems
requirements for the number and quality of the crew. personnel costs increased by 20%. However, we
NAI technology follows current trends in transport. believe that the effect on a global scale will
It enables to manage vessel from the level of single, ultimately lead to a reduction in the cost of crew,
virtual entity. Such management centre can be which at the same time reducing social costs,
located in VTS, where all critical data will be provides a solid base to conclude that such changes
delivered via satellite or other media. VTS becomes are desirable. The authors also do not have doubts
monitoring and supporting centre, while the basic that the money saved in this way, funds partly, if not
crew is still on board. In the further future, when fully, consumed in order to improve system
unmanned vessels become reality, it will be the total reliability, increase bandwidth communications
control and traffic management centre. The biggest systems, and perhaps most importantly, their
challenge that arise that moment are security issues. protection against interference from outside.
Unmanned systems have to be well protected and Technologically, we are ready for these changes. It is
security issues should be as much important as safety the time for next steps. One of them are works on
ones. Performance Standards for Navigation Decision
Support Systems (NDSS) for Collision Avoidance
(CA) initiated by Poland at the 98 session of the
6 FUTURE OF NAUTICAL VESSELS committee MSC.

The autonomous vehicles is the strongest trend of

the development of transport technology in the XXI REFERENCES
century. Maritime transport is probably the most
conservative transport sector, but even this sector Antão, P. & Guedes Soares, C. 2008. Causal factors in
will not avoid general trend and in near future accidents of high-speed craft and conventional ocean-going
vessels. Reliab. Eng. Syst. Saf. 93(9): 1292–1304.
autonomous vessels will dominate on the oceans. Banas, P., Pietrzykowski, Z. & Wójcik, A. 2013. A Model of
The process already starts, as usually first in the Inference Processes in the Automatic ́ Maritime
navy, which apart from small stand-alone objects, Communication System. Communications in Computer and
also operates unmanned destroyers like Sea Hunter Information Science 395: 7–14.
(The Guardian, 2016). There are also few running or Celik, M. & Cebi,S. 2009. Analytical HFACS for investigating
completed project, which aims to develop human errors in shipping accidents. Accid. Anal. Prev.
41(1): 66–75.
autonomous vessel e.g. commercial project led by Harrald, J.R., & et al., 1998. Using system simulation to model
Rolls-Royce or Saab, or research project MUNIN. the impact of human error in a maritime system. Saf. Sci.
From technological point of view, we are ready to 30(1): 235–247.
send an autonomous commercial vessel for example http://www.imo.org/en/About/Events/WorldMaritimeDay/Docu
through the ocean. The biggest concern is avoiding ments/World%20Maritime%20Day%202016%20-
other targets. For this reason anti-collision systems %20Background%20paper%20(EN).pdf
https://www.theguardian.com/us-news/2016/apr/08/us-military-
like NAVDEC were developed. IMO currently christens-self-driving-sea-hunter-warship
works on performance standards for such systems, https://www.gov.uk/government/uploads/system/uploads/attach
which is the first step towards autonomous vessels ment_data/file/9015/seafarer-projections-2011.pdf
(Wołejsza P., et al., 2013). http://www.ect.nl/en/content/ect-delta-terminal
IALA, IALA Technical clarifications on recommendation ITU-
R M. 1371-3, edition 2.4. 2009.
IALA, IALA Recommendation A-126 On The Use of the
7 RESUME OR CONCLUSIONS Automatic Identification System (AIS) in Marine Aids to
Navigation Services Edition 1.5, 2011.
In an era of automation and the rapid development of
autonomous systems, the shipping industry remains
Kulbiej E. 2016. Relevance of the relativistic effects in satellite Wójcik, A., Banas, P. & Pietrzykowski, Z. 2014. The schema
navigation. Scientific Journal of Maritime University of of inference processes in a preliminary identification of
Szczecin 47: 85-90. navigational situation in maritime transport.
Kulbiej E. & Wołejsza P. 2016. An analysis of possibilities Communications in Computer and Information Science 471:
how the collision between m/v 'Baltic Ace' and m/v 'Corvus 304–312.
J' could have been avoided. Annual of Navigation 23: 121- Wołejsza, P. 2013. Functionality of navigation decision
135. supporting system – NAVDEC. Navigational Problems 1:
NAV 59/INF.2. 2013. Development of an e-navigation Strategy 43–46.
Implementation Plan. Report on research project in the field Wołejsza, P. 2014. Navigation decision supporting system
of e-navigation submitted by Poland: IMO. (NAVDEC) – testing in real condition. Annual of
NCSR 2/INF.10. 2015. e-navigation Strategy Implementation Navigation 21: 177–186.
Plan. A study on ship operator centred collision prevention Wołejsza, P., Magaj, J. & Gralak, R. 2013. Navigation
and alarm system submitted by the Republic of Korea: Decision Supporting System (NAVDEC) - testing on full
IMO. mission simulator. Annual of Navigation 20: 149–162.
Pietrzykowski Z., Hatłas P., Wójcik A. & Wołejsza P. 2016. Zhang, J.F., Yang, X. D., Zhang, D., Haugen, S. 2013b. Ship
Subontology of communication in the automation of trajectory control optimization in anti-collision
negotiating processes in maritime navigation. Scientific maneuvering. TransNav 7 (1): 89–93.
Journal of Maritime University of Szczecin 46: 209-216.

View publication stats