953
1 INTRODUCTION
Safe operation of both traditional manned ships and
unmanned or autonomous vessels cannot exist
without appropriate collision avoidance systems
based on reliable safety indicators staying in line with
IMO’s COLREGs rules [1]. To be sure whether a
proposed solution is valuable for the aforementioned
collision avoidance system, first it is necessary to
determine their suitability for currently operated
manned ships. For this purpose, it is important to
design a navigation support tool capable of mapping
the encounter situation, calculating the proposed
safety indicators in the context of a given collision
avoidance situation and presenting the result that can
be interpreted by the navigator.
When safety indicators in ship encounters are
considered, the most obvious choice are Distance at
the Closest Point of Approach (DCPA) and Time to
the Closest Point of Approach (TCPA). They both are
implemented in variety of on-board hardware and
software tools. Sometimes they can be supplemented
by Bow Crossing Range (BCR) or Bow Crossing Time
(BCT). But if one is about to apply some more
sophisticated indicators or metrics like e.g. domain-
based Degree of Domain Violation (DDV) [2], a
problem appears that virtually no tool is available
implementing this indicator. Thus, a necessity arose to
build such a ship traffic monitoring and simulation
tool, having access to AIS data stream and
implementing selected safety measures. The tool
would allow to monitor the encounter situation by
utilizing these measures, thus resulting in improved
situation awareness. That would also make possible to
test some newly designed collision avoidance
algorithms designed for traditional or unmanned
(MASS) vessels. This paper documents exactly such a
traffic simulation tool implemented in Python.
The manuscript is organized as follows. Section 2
presents basic information on AIS as a source of data
Simulation Environment in Python for Ship Encounter
S
ituations
Ł. Stolzmann
1
& J. Szłapczyńska
2
1
Gdynia Maritime University, Gdynia, Poland
2
Gdańsk University of Technology, Gdańsk, Poland
ABSTRACT: To assess the risk of collision in radar navigation distance-based safety measures such as Distance
at the Closest Point of Approach and Time to the Closest Point of Approach are most commonly used. Also Bow
Crossing Range and Bow Crossing Time measures are good complement to the picture of the meeting situation.
When ship safety domain is considered then Degree of Domain Violation and Time to Domain Violation can be
applied. This manuscript provides a description of a ship encounter simulation software tool written in Python
accompanied by a case study, implementing all the measures mentioned above. It offers a radar-like Graphical
User Interface (GUI), is able to track AIS-based traffic or encounter scenarios stored in local files. The tool
features several additional functions e.g. Variable Range Marker (VRM) or Electronic Bearing Line (EBL). The
simulator might be a test sandbox for advanced collision avoidance algorithms.
http://www.transnav.eu
the
International Journal
on Marine Navigation
and Safety of Sea
Transportation
Volume 17
Number 4
December 2023
DOI: 10.12716/1001.17.04.
22
954
of nearby traffic. Section 3 briefly recalls definitions of
basic risk-related metrics applicable in ship encounter
situations. Section 4 introduces the Python simulation
tool including details on its initial assumptions,
applied technologies, GUI design and key features.
The section concludes with possible further
development of the tool. A case study with exemplary
encounter situation analysis with assistance of the tool
is presented in Section 5. The final section concludes
and summarizes the material presented.
2 AIS AS SOURCE OF INFORMATION ON
NEARBY TRAFFIC
Ship situation awareness seems a crucial element of
her safety when maritime traffic is considered. The
situation awareness is especially important for
autonomous vessels. Typically for such marine units,
they can gather information on the other vessels in the
vicinity based on their radar and ARPA displays.
However, due to its analogue roots, radar/ARPA can
suffer e.g. from the shadow effect and are able to
cover a limited range around a ship. Thus, in practice
the most common way of building the situation
awareness for autonomous vessels is by utilization of
Automatic Identification System (AIS).
The AIS is a radio device enabling automatic:
− transmission to suitably equipped shore stations,
other vessels and aircrafts: data identifying the
ship and its type, and specifying its current
position, course, speed, navigation status and
transported dangerous goods, as well as short
safety information,
− receipt of such information from similarly
equipped vessels,
− position monitoring and vessel tracking,
− data exchange with devices ashore.
The AIS device can be installed:
− on ships,
− on the shore as a so-called base and relay device,
− at the centre of the Vessel Traffic Control Service
(VTS),
− on Aids to Navigation (AtoN).
AIS devices are equipped with a Very High
Frequency (VHF) receiver operating on:
− channel 70 using the Digital Selective Calling
(DSC) technique,
− channel 87B or 88B using the Time Division
Multiple Access (TDMA) technique.
AIS utilizes special technique to provide an
algorithm that distributes transmit frames to
individual time devices. The transmission time is
divided into time frames. A single frame lasts one
minute, and its beginning and end coincide with the
beginning and end of the UTC minute.
All receivers must have the same reference time,
because it is used to number the elementary frames in
which these devices work. In the case of a non-
uniform time pattern, there could be a different
numbering of elementary frames, which would cause
transmission conflicts, such as overlapping. This
phenomenon can occur when receiving a weak signal
transmitted by a device at a considerable distance. If
the message transmitted by them is incomplete or
there is a format distortion or data errors, its
transmission is discriminated, i.e. the allocated
elementary frame will be received and the entire
transmission schedule ceases to exist. This frame, if
necessary, is allocated to another AIS having a
stronger signal. In order to eliminate the problems
associated with the reception of weak or distorted
signals from distant devices, messages transmitted by
AIS at distances greater than 100 nautical miles are
automatically discriminated.
Additionally, the following transmission access
methods are distinguished:
− ITDMA - extended TDMA (Incremental Time
Division Multiple Access) system,
− RATDMA - access to TDMA (Random Access
Time Division Multiple Access),
− FATDMA – frequency multiple access (Fixed
Access Time Division Multiple Access),
− SOTDMA - self-organizing time division multiple
access.
Depending on the adopted method of access to
transmission, the following AIS work methods are
distinguished:
− autonomous also called continuous - transmitting
at time intervals:
− static - every 6 minutes and on request,
− dynamic - at intervals depending on the
situation,
− regarding travel - every 6 minutes, on request
and after each change of any data,
− safety information - after entering and on
request.
− assigned mode - the frequency and time moments
of transmission of position reports are determined
automatically by an authorized base device acting
independently or via a so-called relay device (AIS
transponder),
− pooling mode - the ship's AIS transmits the data
upon receipt of an interrogation signal sent by the
AIS of another vessel or aircraft or by the base unit.
As not all vessels operating at sea are subject to the
SOLAS convention [3], the ship AIS equipment has
been divided into two basic classes, namely:
− Class A, intended for sea-going ships, on which
the device is required in accordance with the
provisions of Regulation 19.2.4.1-3 "Carriage
requirements for shipborne navigational systems
and equipment", Chapter V of the 1974 SOLAS
International Convention for the Safety of Life at
Sea.
− Class B, intended for units on which the
equipment does not have to be installed in
accordance with the requirements of the SOLAS
regulation (ships of less than 300 gross tonnage
engaged on international voyages, ships of less
than 500 gross tonnage not engaged on
international voyages, and fishing vessels).
3 RISK-RELATED METRICS IN SHIP ENCOUNTER
SITUATION
When situational awareness is achieved by means of
available AIS data, then risk of possible collision or
damage in an encounter situation can be estimated.
955
Usually, instead of a direct risk estimation, the
navigator utilizes sets of accompanying distance and
time metrics. Among them the most popular are
Distance at the Closest Point of Approach (DCPA)
together with Time to the Closest Point of Approach
(TCPA). When bow distance is taken into account
then one can apply Bow Crossing Range (BCR) and
Bow Crossing Time (BCT). However, when ship
safety domain is considered none of the above metric
applies perfectly. In such situation one can utilize
Degree of Domain Violation (DDV) and Time to
Domain Violation (TDV). The following subsections
briefly recall definitions for all of the abovementioned
metrics.
3.1 DCPA & TCPA
The most popular approach parameters - DCPA and
TCPA have long been present in every radar
equipped with ARPA. To this day, they are an
important component of risk assessment during
seagoing encounters. The advantage is the simplicity
of their determination, but at the cost of the lack of a
perfect representation of the situation. To begin
determining the formulas for DCPA & TCPA, it is
assumed that calculations concerning the above
mentioned are carried out in the Cartesian coordinate
system. This has the following benefits [4]:
− simplicity of the equations of motion,
− reduction of circular and trigonometric functions
that carry some accuracy errors.
Let’s assume that the Cartesian coordinate system
with its own ship in its centre is a movable plane
tangent to the earth's surface. The objects in the area
of interest will be plotted on the adopted system by
means of a geographic projection made with the use
of WGS-84, the result of which will be X coordinates
for longitude and Y for latitude. Thus, let the vector of
true velocity along the x axis be V
tx and along the y
axis V
ty. Similarly with the relative velocity vector: Vrx
for the x axis and V
ry for the y axis.
The relations between the own ship's motion
parameters and the object's motion parameters are as
follows [4]:
22
22
= +
= +
= +
= +
Tx rx x
Ty ry y
T Tx Ty
r rx ry
V VV
V VV
V VV
V VV
where V
x,Vy are own ship movement parameters,
22
ψ
ψ
=
=
= +
x
y
xy
V Vsin
V Vcos
V VV
where ψ is a course over ground of own ship.
It follows from the above:
ψ
ψ
ψ
ψ
= +
= +
= −
= −
Tx rx
Ty ry
rx Tx
ry Tx
V V Vsin
V V Vcos
V V Vsin
V V Vcos
Now, it is possible to apply the following
formulas:
2
−
=
+
= −
ry rx
CPA
r
rx ry
CPA
r
XV YV
D
V
XV YV
T
V
3.2 BCR & BCT
Early 1980s saw the implementation of the BCR &
BCT ship to ship collision risk indicators in radar
systems. BCR stands for Bow Crossing Range and is
understood as the distance at which one ship crosses
ahead of another’s bow (or astern, if negative) [5].
BCT, in turn stands for Bow Crossing Time and is the
time when BCR occurs.
Even though BCR & BCT pair is listed as a safety
indicator that should be used and appropriately
interpreted by the OOW (Officer On the Watch), in
the official, required IMO's Model Course on Radar
Navigation at Operational Level it is only a feature of
INS (Integrated Navigation Systems) that is optional.
Today navigators use it as a secondary indicator of the
type of ship-ship encounter as a supplement to DCPA
safety measure utilization. As shown in Fig 1, BCR &
BCT are used to describe the distance (and time to
reach it) between two ships when they are crossing
one another.
Figure 1 Visual representation of BCR & BCT
On the basis of the distance criterion, a meeting
situation is recognized as an instance of BCR.
Although the general recommendation is to take into
956
account distances not less than 1 NM, the rigorous
range of values, as in the case of DCPA, is not
imposed. However, there is no strict literature
concerning sustained conditions in encounter
situations [6] and the scientific literature contains
diverse values [7].
It is important to recognize that the crossing is a
specific kind of encounter that occurs when one ship
approaches another from the COLREG visibility
sector (regardless if ahead or astern) of just one
sidelight. Head-on encounters and overtaking are not
regarded as instances of the Bow Crossing Range.
Therefore, it becomes clear that BCR might be a useful
measure when neither of the aforementioned
statements is accurate. The BCR & BCT value (positive
or negative) informs the navigator of the ship passing
ahead or astern and their respective COLREG Rule 15
requirements [6].
For BCR & BCT all calculations are made in the
two-dimensional space of Cartesian coordinates:
relative bearing to TS is assumed as
(
) (
) (
)
= −
rel OS true OS OS t
brg brg hdg
where
( )
_ arctan 180
= −°
true OS
Y
brg wrap angle deegres
X
where
(
) (
)
( ) ( )
= −
= −
OSt TSt
OSt TSt
XX X
YY Y
aspect is assumed as
( ) ( )
( ) ( )
360 180
180
−° >°
=
≤°
rel TS rel TS
rel TS rel TS
brg if brg
aspect
brg if brg
where
( ) (
) ( )
= −
rel TS true TS TS t
brg brg hdg
where
( )
_ arctan 180
= −°
true TS
Y
brg wrap angle deegres
X
where
( ) ( )
( ) ( )
= −
= −
TSt OSt
TSt OSt
XX X
YY Y
According to [6] BCR is understood as
( ) (
)
(
)
( )
(
)
_ , , =
bow OS bow OS
BCR nearest points X Y hull TS
therefore
=
TS perspective
SOG
BCR
BCT
TS
.
3.3 DDV & TDV
When distance-based ship safety domain [8] is
considered, neither DCPA & TCPA nor BCR & BCT
are able to provide apt and direct information on
possible domain violation. Thus, in [2] Szlapczynski
and Szlapczynska proposed a brand new pair of
domain-based risk metrics, namely Degree of Domain
Violation (DDV) and Time to Domain Violation
(TDV). They provided there analytical formulae for
DDV & TDV calculation for a standard Coldwell’s
elliptical domain. These metrics have been recently
applied to near-miss analysis and Collision Alert
System frameworks in [9].
In order to calculate values of DDV & TDV for
configurable Coldwell’s domain (with a, b, da and db
parameters) it is assumed that the coordinate system
is a two-dimensional Cartesian one with its own ship
in its centre, as presented in Fig 2. Please note that the
rotation angle α is calculated unlike in sea navigation
i.e. counter clockwise from the X axis.
Figure 2 Assumptions for DDV & TDV: elliptical domain
presented in Cartesian coordinate system with own ship in
its centre
Therefore:
(X, Y) – relative position of a target,
(Xe, Ye) – relative position of the centre of an ellipse
being the target's domain,
Xe = X + h
h= ∆acosα+ ∆bsinα
Ye = Y + k
k= ∆asinα- ∆bcosα
(Vx, Vy) – components of the relative velocity of a
target,
α – the rotation angle of the target's domain (being
equal to course angle of the target), measured
counter clockwise from X axis to the tip of a target's
true speed vector
The elliptic domain moves with the relative speed
of a target:
957
( )
( )
=++
=++
ex
ey
X t X hf V t
Y t Y kf V t
The parametric equation of a rotated ellipse with a
centre in (X
e(t), Ye(t)) as a function of time is:
(
)
( )
( )
( ) ( )
(
)
22
22
1
αα α α
+−
+=
e e ee
X t cos Y t sin X t sin Y t cos
ab
The parametric equation of the f-scaled ellipse
(with the same centre) as a function of time is:
( ) ( )
( )
( )
( ) ( )
( )
( )
22
22
22
1
αα αα
+−
+=
e e ee
X t cos Y t sin X t sin Y t cos
ft a ft b
Solving the latter gives a formula for f(t), as
presented in:
( )
(
)
2
21 22 1 1 1
1,2
2
2
− + ± ++
=
B B t Dt Et F
ft
A
where
22
21 1 1
2
1 22 2 23
1 21 22 2 22
2
1 21 2 21
1
4
24
4
=++−
= −
= −
= −
A A h B hk C k
D B AC
E B B AC
F B AC
where
( ) (
)
21 1 1 1 1
22= +++B AX BY h CY BX k
with
2
2 2 2 2
1,2
2
4
2
−± −
min
E E DF
t
D
,
where
( )
( )
2
2 1 22 1
2
2 1 22 1
22
2 1 22 1
= −
= −
= −
D DB D
E EB D
F FB E
where
( ) ( )
22 1 1 1 1
22=+++
xy yx
B AV BV h CV BV k
whose f(t) minimum over time t is the approach factor
fmin. DDV is then obtained by substituting the fmin to
DDV=max(1-f
min,0). We assume that the following
should be substituted:
22
1
22
1
22
22
1
22
11
2
αα
αα
αα
= +
= −
= +
cos sin
A
ab
B sin cos
ab
sin cos
C
ab
To determine TDV it is necessary to solve
following equations:
22
3 3 33 3 3 33
1
33
22
3 3 33 3 3 33
2
33
44
min ,
22
44
max ,
22
−− − −+ −
=
−− − −+ −
=
B B AC B B AC
t
AA
B B AC B B AC
t
AA
assuming that:
( )
( )
22
31 1 1
31 1 1
22
31 1 1
2
1
=++
= ++ +
= + +−
x xy y
ee ee ey ex
e ee e
A AV BV V CV
B AX V CYV B X V YV
C AX BX Y CY
where t
1<t2,
t
1 – the time remaining to entering the target’s
domain,
t
2 – the time remaining to leaving the target’s domain.
According to [2] there are three possible cases here:
t
1<0 and t2<0 says that a domain has already been
entered and left,
t
1≤0 and t2≥0 says that a domain has already been
entered but not left,
t
1>0 and t2>0 says that a domain will be violated in
time t
1 and left in t2.
4 PYTHON-BASED SHIP ENCOUNTER
SIMULATION ENVIRONMENT
This section describes a simulation tool for ship
encounters implemented using Python language. We
have chosen Python as one of the most popular high
level languages these days, offering a wide range of
built-in functions and easily available open source
modules or packages covering various applications.
Moreover, the tool being described here has been a
part of a wider software and hardware solution, built
in the course of ENDURE project. Thus, a unified
policy towards software implementation was an
important factor here.
Obviously, there are numerous tools available, e.g.
[11-17], offering similar ship traffic simulation or ship
navigational decision support functionality. Some of
the recent ones are also implemented in Python, the
other are written in C/C++, MatLab or other high level
languages. Majority of the solutions are not in the
public domain nor have open source licence and even
if they are open sourced, they have got a non-
compatible functional range [11]
. Thus, it was
necessary to design and implement our own tool,
958
fully customizable, offering the exact functionality
that was required in the course of the ENDURE
project.
The tool described here is able to monitor live
encounter traffic situation AIS data. It is also possible
to utilize offline situation data stored in a local file.
Moreover, the tool implements the following risk-
related measures: TCPA, DCPA, BCR, BCT, DDV and
TDV, their values can be monitored throughout the
encounter. The following subsections present initial
assumptions of the tool, the applied technologies, GUI
and key features of the tool and finally directions for
further tool development.
4.1 Initial assumptions
The application was designed with utilization of
Model-View-Controller design pattern. The model
consists of the application operation logic that
interacts with an external source of information (AIS,
database). It includes functionality responsible for
data management and processing. All data that will
be presented to the user is contained here. The view
consists of all interactive and non-interactive objects
that will be displayed to the user. It is a set of features
responsible for the visualization of data managed by
the model. The controller is a set of features
responsible for intermediation between the view and
the model. This is where events are captured and
carried out.
The application can be autonomous, provided that
it receives AIS reports. In the case of this software,
user intervention is not obligatory for the program to
function. The role of the program operator can only be
based on observing a self-updating decision support
program. The program allows the user to customize
the way information is displayed by changing the
display, and to use additional tools to facilitate
navigation.
All the presented algorithms assume that vessels
are in sight of one another and no other special
circumstances (e.g. restricted visibility) apply.
4.2 Applied technologies
The software was developed using Python version 3.
The pygame graphic module was used which made
graphic visualization possible, the module is a set of
Python submodules designed for writing video
games. The vast majority of the simulation tool was
written from the scratch using pure Python. However,
the need to use the following python modules turned
out to be indispensable:
− pyais for encoding and decoding AIS message,
− pandas for data analysis and manipulation,
− numpy for arrays handling and linear algebra,
− pyproj for cartographic projections and coordinate
transformations,
− geopy for calculation of geographic distances.
4.3 GUI and key features of the application
The software is a window application with a user
interface operated by the mouse. The application
interface was inspired by the interfaces of radar
devices and ECDIS. The advantage of this solution is a
relatively intuitive and easy-to-use interface for
navigators.
Figure 3 shows an overview of the program’s GUI.
In the central part, on the azimuth dial, target vessels
visible in a given range are displayed.
Figure 3. Overview of the program’s GUI
The range and the orientation as well as motion
mode are configurable parameters. The range can be
set between 0.75 and 48 NM, motion mode parameter
offers relative or true motion options, while the
orientation can be north up or head up. By default, the
range is set to 12 miles and displayed in relative
motion north up as this is the IMO recommended
setup for collision avoidance purposes.
As shown in Figure 4 it is possible to use in this
tool Variable Range Markers (VRM) and Electronic
Bearing Lines (EBL) simultaneously. They provide the
ability to quickly measure bearing and distance to any
visible object.
Figure 4. Usage of Variable Range Markers (VRM) and
Electronic Bearing Lines (EBL)
Variable range marker is a navigational aid for the
operator. After clicking the VRM button it displays a
circle that allows to quickly measure the distance to
an object without the need to take it for an acquisition.
The electronic bearing line activated with the button
indicated in the Figure 4 is another tool supporting
the navigation process, which allows, similarly to
VRM, without the need to acquire a target ship, to
determine the bearing to any object.
959
Figure 5. Range rings, true motion mode
Similarly to radars, the option of displaying
distance circles has also been introduced. As
presented in Figure 5, the range rings are drawn after
pressing the RINGS button. The distance rings enable
better orientation of the spatial position of objects,
because they divide the range of the displayed area
into six equal parts.
Figure 6. Acquisition of fully initialized target vessel
In this software, target ship acquisition is done
differently from the ARPA equipped radar. Since the
source of the data is AIS, and not the radar pulse as in
the case of a radar, it is not necessary to wait a certain
amount of time for the target echo to start being
tracked. The information is available immediately
after clicking the cursor on the desired triangle
symbolizing the target ship. To be precise, it is almost
available because, as described in section 2, the target
ship data is transmitted in two separate reports. One
relates to the dynamic information of the ship, the
other to static information, transmitted less
frequently. For this reason, ships that have not
received a matching type 5 report containing static
data will be referred to as not fully initialized and will
be marked with a white circle in the centre of their
triangle symbolizing the ship’s abstract silhouette.
Although the vessel is not fully initialized, indicators
such as DCPA & TCPA and BCR & BCT are
calculated. The difference between fully initialized
ships and those without a static data report received is
shown in the Figures 6 and 8.
For a fully initialized target, besides DCPA/TCPA
& BCR/BCT, the software is possible to calculate DDV
& TDV as this requires the length of the target ship,
which can be calculated on the basis of the
information contained in the type 4 report.
Additionally, the CADCA (Collision Avoidance
Dynamic Critical Area), presented in the Figure 7, can
be drawn for a fully initialized vessel. CADCA [10] is
a deterministically defined area that geometrically
limits the manoeuvring area of the vessel for which it
is designated. The CADCA shape and size depends
on the movement parameters of the vessel such as
rudder angle, initial forward speed, or planned
alteration of the course. The main assumption of the
CADCA concept is to support navigational decisions
in collision situations. For a detailed description,
please refer to the article [10].
Figure 7. Collision Avoidance Dynamic Critical Area
(CADCA)
Figure 8. Acquisition of partially initialized target ship
Figure 9. Trial manoeuvre
The situation in which a ship is taken for the
acquisition that is not fully initialized (with a white
960
circle on a triangle) is shown in the Figure 8. In this
case, the calculations of the DDV & TDV safety
indicators are omitted due to the lack of all data. In
this situation, however, it is suggested to rely on
DCPA & TCPA until report of type 4 is received and
consequently the required safety indicators are
calculated.
The tool offers the trial manoeuvre calculations for
planning the collision avoidance manoeuvre. After
clicking the TRIAL button, three controls appear to
allow the user to change: course, heading and speed.
The word TRIAL appears on the screen, which
indicates that all changes to the parameters of the own
ship's movements are carried out on a trial basis,
without actually changing them. The functionality of
the trial manoeuvre here consists in presenting the
changes in the relative motion vectors of the target
ship and the change in the value of safety indicators
depending on the planned course, heading or speed
change. The manoeuvre has no time delay, all
planning is done live, the own vessel and target
vessels continue to move during the planning process,
according to the received AIS reports. Figure 9 shows
an example of a trial manoeuvre. The own ship as
well as the target ship moves at a given speed and
heading. After the planned increase of the own ship's
speed, the appearance of the vector of the relative
speed of the target ship in relation to the own ship, in
the form of a cyan line, symbolizing future positions
in given moments of time, is observed.
Figure 10. Filtering of target vessels based on DCPA value
The software features a function of filtering target
ships due to their potential collision risk. The filter is
based on the DCPA value - when this value drops
below 1 NM, the target ship against which the DCPA
is measured is considered as potentially dangerous. In
the filtering mode, ships that do not pose a threat are
marked in grey, while potentially dangerous ships
appear in magenta as in the Figure 10. When filtering
is disabled, all ships are treated as potentially
dangerous and display magenta coloured by default.
When filtering is enabled, all ships that do not
meet DCPA < 1NM are not considered as dangerous
ships. However, it should be considered that the ships
currently not posing a threat may pose a potential
threat in the future - then the program would catch
this and update the information displayed. The value
of the filter is of course customizable.
4.4 Possible further development of the tool
Further possible application development includes
implementation of the functionality enabling the
visualization of electronic maps. This would allow the
program to be used in difficult navigational areas
without the risk of a collision with a stationary object.
Considering the development of the maritime
industry, it is possible to adapt the existing software
functionality to the requirements of collision
avoidance systems of unmanned ships. However, this
is a topic for the long run.
An interesting aspect of the software is its open
architecture. It allows the application to be used as a
test environment enabling relatively simple
implementation of, for example, experimental safety
indicators. A potential prospect for the development
of the application may also be the use of the current
operating logic to create a proprietary functionality
consisting in the analysis of the movement of ships
using AIS. There are many possibilities, but the issue
of their use in the era of rapid scientific and industrial
development is a topic for a separate discussion. At
present, the program may turn out to be useful as an
experimental decision support system, which may be
of interest to the scientific community, the maritime
industry as well as enthusiasts of new technologies.
5 CASE STUDY – ANALYSIS OF A CROSSING
ENCOUNTER BY THE SIMULATOR TOOL
This case study presents an encounter of crossing
from the port side. According to COLREG Rule 15
when two power-driven vessels are crossing so as to
involve risk of collision, the vessel which has the other
on her own starboard side shall keep out of the way
and shall, if the circumstances of the case admit, avoid
crossing ahead of the other vessel [1]. The encounter
situation, recorded in the Stavanger area of the North
Sea is depicted in Figure 11. Duration of the entire
recorded meeting is about 26 minutes. The encounter
took place on November 7th, 2022 at 18:08 CET. The
meeting parameters of the vessels involved in the
considered situation are presented in Table 1. As for
the ship domain Coldwell’s one is assumed with
parameters: a=0.794NM, b=0.397NM, da=0.198NM
and db=0.099NM.
Figure 11. Case study – situation overview: crossing from
starboard
961
According to the COLREGs, the give-way vessel is
the target ship, while the stand-on vessel is the own
ship. The situation requires that the target ship keeps
clear of the own ship by making a clear course
manoeuvre to starboard. However, no such course
change took place in the first 6 minutes of encounter.
Thereafter, minimal course changes are made, which
contradicts the COLREG recommendation to use
significant course changes. As shown in Figure 12, the
DCPA(t) graph indicates that the vessels have kept a
minimum distance of 0.5 NM from each other during
entire meeting. The upward trend of DCPA indicates
that the target vessel was performing a speed increase
manoeuvre. The effect of this manoeuvre become
clearly visible from about 14 minutes. In comparison
with DCPA as the main predictor of collision risk, the
confidence that the risk decreases was obtained after
about 14 minutes - at this point the distance at the
closest point of approach (DCPA) grows
exponentially. Considering the graph of the BCR(t),
shown in Figure 13, the target vessel crossed the
course of the own vessel at a distance of 1.6 NM.
Table 1. Case study – parameters of the ships in meeting
________________________________________________
Own ship Target ship
________________________________________________
Name m/t “Bit Power” “Viking Prince”
Length 116.9 m 89.6 m
Beam 18.0 m 21.0 m
Initial coordinates 59° 3' 18'' N 59° 5' 55.0608'' N
004° 10' 44.112'' E 004° 8' 37.1544'' E
Final coordinates 59° 7' 15.6036'' N 59° 5' 7.1736'' N
004° 10' 41.7'' E 004° 16' 36.768'' E
Initial COG 358.0° 101.0°
Initial SOG 9.8 kn 11.5 kn
________________________________________________
The conclusion resulting from the comparison of
the two safety indicators so far is that used
individually, they are a poor way to assess the risk of
a collision. It can be safely said that using only BCR it
is possible to determine only the type of meeting
situation, and in a rather vague way. On the other
hand, using only DCPA it is possible to ascertain only
the effect of the situation. Combining these two
indicators together, we get a more complete picture of
the situation, including the type of encounter situation
and the effect that will be achieved with the current
movement parameters.
Figure 12. Case study – DCPA values during the encounter
Figure 13. Case study – BCR values during the encounter
Figure 14. Case study – DDV values during the encounter
Slightly different tendency is presented by the
DDV indicator, depicted in Figure 14. It is noticeable
that DDV measure provides useful information in less
time than the previously considered DCPA and BCR.
It was also observed that this indicator was more
sensitive to changes in the parameters of the ship's
motion and its position. It is particularly visible in the
time interval 0 - 2.5 minutes, where the shape of the
curve indicates a sudden change in the nature of the
situation and the successive decrease in the risk of
collision (the Degree of Domain Violation decreases).
In this scenario in case of DCPA, the reliable
collision risk assessment was possible after 12
minutes. In the case of BCR – it was possible after
approximately 8 minutes. In this particular scenario
based on DDV one is able to assess the collision risk in
no more than 4.5 minutes. As depicted in Figure 14,
there is an immediate suggestion (at the start of the
recorded situation) of a DDV value of around 0.5,
which shows a severe domain violation, moreover
with upward trend of DDV values during first 2 min.
of the encounter. After reaching the value of 0.9 in
about 3 minutes of observation, the situation
improves until reaching the value of DDV equal to 0,
meaning no domain violation. Thus, one might
conclude that relying on the DDV value alone in the
risk assessment, the situation in this case was safely
cleared up after approximately 4.5 minutes.
962
For the examined crossing case, for which the
situation allowed the use of DCPA and BCR
indicators for comparison purposes, it is clear that
DDV proves its superiority. The DDV answers the
question of whether there is a risk of collision not only
in a binary way, but, when the risk does exist,
provides also the risk magnitude.
However, the use of the DDV indicator is not
without its drawbacks. Application effectiveness of
metrics like DDV or BCR is indirectly limited by the
use of AIS as its base source of data. One of the
serious limitation here is the specification of the AIS
itself. For example, the user is forced to wait for a
static data report because the length of the target
vessel is required for the DDV or BCR calculation. It is
not uncommon that during a collision avoidance
manoeuvre every minute saved might be crucial, thus
waiting for an AIS report with static data in a critical
situation may end up at best with a conflict with
COLREG Rule 8. The conclusion is that any ship-
length dependant indicator, as e.g. the DDV or BCR
metrics, should be used with caution in real-time
collision avoidance systems.
6 SUMMARY
The presented here ship encounter simulation and
traffic monitoring tool offers features allowing for
easy customization and making possible to test
various collision avoidance solution within its
graphical environment. It is worth noticing that the
tool implements a number of safety indicators,
namely DCPA, TCPA, BCR, BCT, DDV and TDV and
is ready for implementation any new metric, if such
necessity arise. As presented in the previous section,
the tool has been validated on live AIS data stream
and real ship encounter scenario. It is planned to
continue development of the tool towards integration
with the s-57 map and/or with the s-100 map. Also it
seems promising to extend the tool in future by
including weather forecast, hydrographic information
and ship stability decision support.
ACKNOWLEDGEMENTS
The study described has been performed as part of the
Detection, prediction, and solutions for safe operations of
MASS (ENDURE) project (number
NOR/POLNOR/ENDURE/0019/2019–00), supported by the
Polish National Centre for Research and Development and
financed by Research Council of Norway.
REFERENCES
[1] IMO, “Convention on the International Regulations for
Preventing Collisions at Sea,” COLREG.
[2] R. Szlapczynski and J. Szlapczynska, “An analysis of
domain-based ship collision risk parameters,” Ocean
Eng., vol. 126, pp. 47–56, 2016, doi:
10.1016/j.oceaneng.2016.08.030.
[3] “SOLAS Convention.”
https://www.imo.org/en/About/Conventions/Pages/Inter
national-Convention-for-the-Safety-of-Life-at-Sea-
(SOLAS),-1974.aspx
[4] A. Lenart, “Approach Parameters in Marine Navigation
– Graphical Interpretations,” TransNav, Int. J. Mar.
Navig. Saf. Sea Transp., vol. 11, no. 3, pp. 521–529, 2017,
doi: 10.12716/1001.11.03.19.
[5] Ł. Stolzmann, “Bow Crossing Range Correlation of Small
Vessels - Ais Data Analysis With Prospective
Application To Autonomous Ships.,” Sci. J. Gdynia
Maritme Univ., vol. 6988, no. 121, pp. 41–52, 2022, doi:
10.26408/121.04.
[6] M. Gil, P. Kozioł, K. Wróbel, and J. Montewka, “Know
your safety indicator – A determination of merchant
vessels Bow Crossing Range based on big data
analytics,” Reliab. Eng. Syst. Saf., vol. 220, 2022, doi:
10.1016/j.ress.2021.108311.
[7] A Guide to the Collision Avoidance Rules. Elsevier, 2012.
doi: 10.1016/C2010-0-68322-2.
[8] R. Szlapczynski and J. Szlapczynska, “Review of ship
safety domains: Models and applications,” Ocean Eng.,
vol. 145, no. January, pp. 277–289, 2017, doi:
10.1016/j.oceaneng.2017.09.020.
[9] R. Szlapczynski and J. Szlapczynska, “A ship domain-
based model of collision risk for near-miss detection and
Collision Alert Systems,” Reliab. Eng. Syst. Saf., vol. 214,
no. May, p. 107766, 2021, doi: 10.1016/j.ress.2021.107766.
[10] M. Gil, “A concept of critical safety area applicable for
an obstacle-avoidance process for manned and
autonomous ships,” Reliab. Eng. Syst. Saf., vol. 214, p.
107806, 2021, doi: 10.1016/j.ress.2021.107806.
[11] S. Blindheim, T. A. Johansen, "Electronic Navigational
Charts for Visualization, Simulation, and Autonomous
Ship Control," IEEE Access, vol. 10, pp. 3716-3737, 2022,
doi: 10.1109/ACCESS.2021.3139767.
[12] Z. Pietrzykowski, P. Wołejsza, Ł. Nozdrzykowski, P.
Borkowski, P. Banaś, J. Magaj, J. Chomski, M. Mąka, S.
Mielniczuk, A. Pańka, P. Hatłas-Sowińska, E. Kulbiej, M.
Nozdrzykowska “The autonomous navigation system of
a sea-going vessel” Ocean Engineering, Vol. 261, 2022,
112104, https://doi.org/10.1016/j.oceaneng.2022.112104.
[13] P. Wołejsza, J. Magaj, R. Gralak “Navigation Decision
Supporting System (NAVDEC) — testing on full mission
simulator” Annual of Navigation 20(1), 2013,
https://doi.org/10.2478/aon-2013-0010.
[14]
J. Lisowski, "Dynamic Games Methods in Synthesis of
Safe Ship Control Algorithms", Journal of Advanced
Transportation, vol. 2018, Article ID 7586496, 8 pages,
2018. https://doi.org/10.1155/2018/7586496.
[15] A. Lazarowska,”Evaluation of a deterministic real-time
path planning algorithm for autonomous ships using
radar data” Procedia Computer Science, Vol. 192, 2021,
pp. 863-872, https://doi.org/10.1016/j.procs.2021.08.089.
[16] Ł. Kuczkowski, R. Śmierzchalski, (2017). Path planning
algorithm for ship collisions avoidance in environment
with changing strategy of dynamic obstacles. In:
Mitkowski, W., Kacprzyk, J., Oprzędkiewicz, K., Skruch,
P. (eds) Trends in Advanced Intelligent Control,
Optimization and Automation. KKA 2017. Advances in
Intelligent Systems and Computing, vol 577. Springer,
Cham. https://doi.org/10.1007/978-3-319-60699-6_62.
[17] Y. Huang, L. Chen, R. R. Negenborn, P.H.A.J.M. van
Gelder “A ship collision avoidance system for human-
machine cooperation during collision avoidance” Ocean
Engineering, Vol. 217, 2020,
https://doi.org/10.1016/j.oceaneng.2020.107913.
