347
1 INTRODUCTION
From the beginning of the introduction of radars on
vessels, their primary application was the detection of
objects in the area, primarily to warn of a possible
threat. The greatest benefit of implementing radars on
civilian vessels after World War II was to obtain more
accurate information for collision avoidance, although
radar can also be used for positioning a vessel.
However, the use of radar on a moving vessel implies
the need to take her movement into account, because
the interpretation of the image is not very easy. For
many years the radar image has stabilized with the use
of a gyrocompass, and additionally with the use of a
log [1].
The gyrocompass is mentioned above as the source
of heading data as it is the most common solution,
however, a magnetic compass equipped with special
sensors can be used as well. This is a separate issue,
because currently there are many different design
solutions in gyroscopic technology [2], which entails a
different specificity of the oscillation of the course
indications, which are natural for gyrocompasses. The
use of a magnetic compass with the option of data
transmission entails the problem of taking declination
and deviation into account, although many modern
Is It Permissible to Use GPS Data to Avoid Collisions?
A. Felski & K. Jaskólski
Polish Naval Academy, Gdynia, Poland
ABSTRACT: Automatic Radar with Plotting Aids is the basic means of preventing collisions at sea for many years.
However, the use of the radar on a moving vessel requires image stabilization, which has been at least for the last
50 years solved by coupling with the gyrocompass and the log. In the present century, the widespread use of
Global Navigation Satellite System receivers has led to the common practice of interconnecting this receiver with
many other systems on ships. This is often also the case for radar, although GNSS gives information about
movement related to the ground, whereas the International Maritime Organization recommends using
parameters relating to water. The mandatory and widespread equipping ships with the Automatic Identification
System means that this system is increasingly used in the process of collision avoidance, but also with the use of
ground-referenced data. The aim of the paper is to investigate whether this is acceptable and what are the limits
of this practice. This question becomes increasingly important in the context of the growing number of unmanned
vessels. Not all, especially small autonomous surface vehicles will be equipped with radar and may also use AIS
transmissions in collision avoidance algorithms.
Studies have shown that this may pose a risk of collision. At low ship speeds, if the current speed exceeds 5 knots
and the direction of the current significantly deviates from the course of one of the ships, there is a risk that the
planned maneuver will not be carried out. This may mean that the closest approach distance will be significantly
different from the planned one.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 19
Number 2
June 2025
DOI: 10.12716/1001.19.02.01
348
designs provide the possibility of automatic solution to
this problem. According to many analyses published,
for example, by [3], a noticeable contribution to the
causes of marine collisions comes from the compass.
The authors describe their analyses of marine accidents
by pointing out that about 10% of collisions resulted
from improper operation of vessel compasses.
However, this is a deeper problem, because the
incorrect operation of the compass can affect the poor
operation of the radar or Automatic Radar with
Plotting Aid (ARPA), but also the operation of the
Electronic Chart Display and Information System
(ECDIS) or autopilot, or possibly the work of the officer
of the watch or the pilot.
The use of the radar to the avoidance of collisions is
a well-known problem and widely described in the
literature. Obligatory introduction of the radar with the
option of the automatic tracking of targets (ARPA)
changed a lot on this field, however still this is the
essential problem from the point of view of the safety
at sea and are all the time performed much research in
this area. The perfect review of methods of the
avoidance of the collisions is given in [4].
The problem can be divided into two aspects:
collision risk assessment (conflict detection) and the
planning of a safety manoeuvre (trajectory planning).
These challenges can be addressed using both
kinematic and dynamic methods. In many publications
risk assessment of the collision is performed based on
so-called ship’s domain area around the own ship
which is forbidden for the encountered vessels. Other
methods are also considered, such as the Dijkstra
algorithm, the Artificial Potential Field or A*, but the
main difficulty is considering the international
collision avoidance rules (COLREGS) [5]. However, in
practice more classical methods, based on vector
calculus is still in use.
The introduction of the automatic identification
system (AIS) on ships altered this landscape by
providing a new source of information about the
movement of encountered vessel, however this caused
the new problem, is which the fusion of the
information from two sources [6][7]. This new solution
introduced the new trend in the form of the intelligent,
decision-making supporting systems [8][9][10].
In turn the common use of the Global Navigation
Satellite System (GNSS) on modern vessels has created
a completely different situation. This system,
commonly interpreted as a positioning system, also
provides information about the movement of the object
relative to the sea bottom (over the ground). In
conjunction with the obvious possibility of automatic
transmission of this data in a digital version, the radar
is more and more commonly combined with a GNSS
receiver and the course over the ground (COG) and
speed over the ground (SOG) are transmitted. This is
particularly important in relation to the ARPA, which
is mandatory equipment for vessels as the basic system
supporting collision avoidance. This seems to be an
obvious solution in the face of the undoubted
appearance of unmanned vessels, in the transport
variant Maritime Autonomous Surface Ships (MASS)
or the increasingly common surface drones
(Autonomous Surface Vehicle ASV), which are
already eagerly used today, especially for various
measurements and research purposes. Undoubtedly,
the creators of such systems will face the challenge of
implementing several very vague convention rules
[11][12][13][14], using phrases such as safe speed, good
practices, limited trust, etc., combining them with
machines operating on similar principles to ARPA or
AIS. However, this article, will only focus on the issue
of using information about the own vessel's movement
in relation to the sea bottom (Speed Over the Ground
SOG and Course Over the Ground - COG). The
purpose of the considerations described here is to
explain how large errors are introduced by replacing
information about the movement through the water
with information about the movement overground.
This is indirectly related to the AIS system, which is
also more and more commonly used to solve anti-
collision problems and can be the main source of
information on ASV. Theoretically, in this system both
COG and SOG are transmitted, as well as course
(heading - HDT) and speed through water (STW).
However, the author's observations show that much
more attention is paid to COG and SOG than to the
other two components of the movement. The
published results of research into the reliability of data
transmitted via AIS prove that the reliability of HDT
provided by this system is significantly lower than all
other data [15] [16].
Since calculating anti-collision maneuvers require
knowledge of one's own ship's course and speed, until
the end of the 20th century it was obvious that deck
officers (Officer of the Watch OOW) would use the
gyrocompass and water speed log for this purpose, for
the simple reason that these means had been widely
used. However, when considering the issue of speed
measurement, a much wider range of solutions is
available and, depending on the meters used, the speed
of the vessel can be STW or SOG. In addition, this
information can be presented as two components of
motion in the plane (forward/back, along the vessel's
main axis and lateral speed) or only as a vector in main
axis (for or aft direction). In this context, it should be
noted that the International Maritime Organization
(IMO) Resolution MSC 334(92) [17] specifies as follows:
1.1. Devices to measure and indicate speed and
distance are intended for general navigational and
vessel maneuvering use. The minimum
requirement is to provide information on the
distance run and the forward speed of the vessel
through the water or over the ground. Additional
information on the vessel’s motions other than on
the forward axis may be provided. The equipment
should comply fully with its performance standard
at forward speeds up to the maximum speed of the
vessel. Devices measuring speed and distance
through the water should meet the performance
standard in water of depth greater than 3 m beneath
the keel. Devices measuring speed and distance
over the ground should meet the performance
standard in water of depth greater than 2 m beneath
the keel.
1.2. Radar plotting aids/track control equipment
requires a device capable of providing speed
through the water in the fore and aft direction.
This recommendation is often not respected, and
the aim of further research is to analyze the possible
effects of replacement of the HDT and STW by COG
and SOG. A thorough analysis of world literature does
not provide knowledge of any research on this topic. In
349
the further part of the article, the essence of the
differences between the use of COG/SOG or course
through water CTW/STW in planning a maneuver and
the graphical analysis of three cases and the synthesis
of the issue based on experiments having been carried
out on the simulator are presented.
As the literature does not provide any information
on the impact of the use of COG/SOG data on the
effectiveness of anti-collision maneuvers, an attempt
was made to estimate this factor. The aim of the
presented research is to assess this impact, assuming
that it is important for system designers to prevent
collisions of autonomous ships (both MASS and ASV).
This is a related issue that would require separate
research, as it appears particularly important in the
context of collision avoidance for small Autonomous
Surface Vehicles (ASVs), which may not be equipped
with radar and will rely solely on AIS-type information
exchange. In this article, we did not address these
concerns, as the AIS system provides data about the
movement of the vessel carrying the AIS device,
eliminating the need for additional analysis of the
relative movement of the encountered ship.
Unfortunately, the available publications do not
address this type of problem. Efforts to combine ARPA
and AIS data, often alongside the Electronic Chard
Display and Information System (ECDIS), focus mainly
on integrating data with different characteristics. These
differences included lack of time synchronisation,
varying dimensions, and differing levels of accuracy
(see, for example [6]). Such efforts are also aimed at
planning safe manoeuvres using the “ship domain”
concept when multiple vessels are nearby. These plans
generally assume that the manoeuvre will be executed
precisely.
The issue discussed here can, in fact, be viewed
because of human error on the vessel’s bridge. If the
Officer of the Watch (OOW) calculates a safe COG, they
must ensure the vessel follows this course. However,
the COG is influences by sea currents. To achieve this,
simply adjusting the heading (HDT), or the ship’s
orientation relative to the north, is insufficient. Doing
so requires additional calculations to account for the
effect of a current at a particular direction.
Further considerations are carried out with the
assumption that the own ship monitors the relative
motion of the encountered ship and, on this basis,
assesses the movement of the encountered ship and
then plans a maneuver to avoid a collision. However,
maneuver planning based on data transmitted via AIS
can lead to similar effects, if the data relates to the
ground. In the context of ASV, interesting conclusions
are provided in publications [4]and [11].
2 COLLISION AVOIDANCE BASED ON MOTION
VECTOR OVER THE GROUND
As mentioned earlier, the widespread use of the GNSS
on a modern vessel and the resulting possibility of
automatic data transmission between devices leads to
the situation that common practice is to combine a GPS
receiver with an ARPA and automatically transmit
data about the COG and SOG of the own vessel, which
is contrary to the IMO recommendations. For radar
image stabilization, such a solution does not raise any
objections, but the question arises what impact it may
have on the effectiveness of collision avoidance for the
work of ARPA or similar devices? The difference
between the motion vector relative to the water and
the ground is self-evident for mariners. The influence
of the wind varies more than the influence of the
currents, because it results from the hull structure, the
direction of the wind relative to the vessel, and will
most likely be different for the own vessel and the
vessel encountered. It will also change as a function of
the wind direction. On the other hand, we can assume
that the movement of water masses within the radius
of operation of a typical navigational radar, i.e. within
the radius of around 20 nautical miles (NM), will
probably be the same for both vessels.
Let us consider what differences will appear in the
results of the process of determining the motion vector
of the encountered vessel and what impact it may have
on the implementation of the maneuver if a GNSS
receiver is connected to ARPA as a source of
information about the course and speed. Let us assume
that the considerations refer to the open sea, so there is
no need to take any additional restrictions relevant to
maneuvers into account.
This issue can be considered by relating to the
motion of both vessels to an Earth coordinate system,
but navigators more commonly analyze it by relating
the movement of the target to the movement of their
own vessel. Of course, one can use analytical relations,
as was done e.g. in [18] but in the rest of the article the
vector calculus suggested in most navigation textbooks
and manuals, e.g. [19] [20], will be used.
Figure 1 shows a typical situation where a target
vessel encountered at points A and B was observed
within a few minutes, and the analysis of her relative
motion suggests that it poses a collision hazard (solid
line in red is passing too close to own ship in the center
of the radar picture).
In this case, according to COLREG Rule 15, when
two power-driven vessels are so positioned relative to
each other that there is risk of collision, the vessel
which has the other vessel on her starboard side shall
keep out the way. On this figure the motion vectors of
the own vessel are shown in blue, the sea current vector
in green, and the motion vector of the encountered
vessel is red. Whereby the solid line represents
movement over the ground and the dashed line
represents movement through the water. The example
confirms the obvious fact that by relating the relative
motion of the encountered vessel to the vector of the
own vessel over the ground, we will obtain
information about the motion of the encountered
vessel over the ground, and assuming the own motion
through the water, we will receive information about
the encountered vessel motion through the water.
Having information about the real motion of the
encountered vessel, it is of course possible to plan a
collision avoidance maneuver. On this basis we assume
that a result of such a maneuver, the motion of the
encountered vessel in relation to the own vessel must
ensure the appropriate distance of the closest point of
approach (CPA) for both vessels. It is not important
whether ARPA or any other instrument or method has
been used for this purpose. It is important that the
result of the maneuver planned in this way is
information on what the own vessel's motion vector
350
should be to ensure avoidance of a collision. In other
words, if the officer of the watch (OOW) uses ARPA
connected to the GPS, he will get information about
what COG and SOG will help avoid a collision.
Undoubtedly, the situation will look similar on an
unmanned vessel when the appropriate automaton
performs this task.
Figure 1. Determination of the motion vector of the
vessel encountered.
However, the key question is whether, knowing the
value of own SOG and COG, required for avoiding a
collision, the OOW or automatic machine will calculate
the new STW and the new CTW, considering the
current at that moment?
Let us note that in such a case most of OOWs will
probably not carry out additional calculations, which is
seldom noticed. Most often, OOW will change the
course (HDT, not COG) by the number of degrees
indicated, usually without changing the speed. It can
therefore be assumed that, for example, he decides that
a 30-degree turn to starboard should be made by
changing the COG, he will in fact to change the course
30 degrees to the starboard by changing the HDT!
Of course, depending on the distance between the
ships, the relation between the speed of the own vessel
and the target and the speed of the current, as well as
the relation between the course of the own vessel, the
course of the encountered vessel and the direction of
the current, the results of such conduct will be
different. It can be assumed that if the speed of the
current is small relative to the speed of both vessels, the
effect will be small. And if the speed of the current is
higher, the impact may be significant. The question
arises as to what significant changes this approach
causes and where are the limits of acceptability of such
a practice?
Taking the problem in general, it would be
necessary to consider different proportions of the
speed of both vessels and the speed of the current,
different courses of vessels and different directions of
the current, and sectors in which vessels move relative
to each other. This makes the analysis very complex
and multifaceted. Therefore, the following section is
limited to cases where the own vessel should give way,
and its speed and course over the ground are
unchanged. It has been assumed that the target and
own vessel are both a typical merchant, power driven
vessels, and therefore their speed are within 10 to 25
knots. The influence of different current directions at a
constant value of its speed, which is up to 50% of the
SOG of the own vessel, has been analyzed.
3 RESULTS OF INVESTIGATIONS
The issue under consideration is a function of many
factors defining the mutual movement of both ships. It
also depends on environmental conditions, such as the
extent of the water area (possible restrictions on
maneuvers), the direction and strength of the wind and
the type of ship. However, I omitted these factors,
treating the reported research as a rough examination
of the issue. Moreover, considering all the factors
would make it difficult to present the results
synthetically, and the aim is to estimate how big an
impact the analyzed approach has and how
complicated it is. The next two figures show a similar
case, but the difference is in the direction of the current.
This shows how important this information is and how
different the results of the maneuver can be.
3.1 Scenario #1; Target at an Angle of 45o, Sea Current
in parallel with own COG
In Fig. 2 shows a situation when the own vessel must
have to give way to a vessel visible from starboard side
at a relative bearing of approximately 45°o. It has been
assumed that both vessels are affected by a current
with a direction consistent with the COG of the own
vessel and a speed slightly less than half the SOG of the
own vessel. This means that the own vessel's CTW is
consistent with the COG and its speed STW is
noticeably lower than the SOG. The analysis of the
situation suggests that the encountered vessel will pass
a short distance astern. This analysis also shows that
the safe passing distance (DCPA) will be ensured by a
turn to starboard of 25° (Fig.2a, dashes red line).
However, to perform such a maneuver, a new heading
and perhaps also a new STW should be calculated.
Such calculations are generally not carried out on the
bridge and most likely a routine turn to the starboard
will be made by the mentioned angle of 25° (Fig. 2b
vector in blue, dashed line). Such a change of the own
course, considering the current, will cause the own
vessel to move on the COG of 015°, not 025°, and her
SOG will drop. In the consequence they will pass closer
than planned (dotted line in red).
Figure 2a. An illustration of the collision avoidance process:
a - planning the maneuver - make a turn to the starboard so
that the COG increases by 25°.
351
Figure 2b. An illustration of the collision avoidance process:
b starboard turn (HDT) of 25° gives the new COG/SOG but
does not give the planned CPA distance.
3.2 Scenario #2; Target at an Angle of 45o, Sea Current
in Parallel the Target’s COG
Let's consider a similar case, assuming the current
velocity remains as in the previous example, but its
direction is consistent with the target's COG,
approximately deflected to the left from its own COG
by 45°. Of course, if the course and speed over the
ground of the own vessel do not change, this means
different parameters of the motion of both the
encountered vessel and the own vessel through the
water, but since the motion over the ground of both
vessels remains the same as in the previous example, a
turn to the starboard should be made again to avoid a
collision by 25° (Fig.3a), which would require a change
in course and possibly speed.
Figure 3a. Illustration of the collision avoidance process
when the current direction is in line with the target COG: a -
planning maneuver, as the movement of both vessels relative
to the ground has not changed (compared the previous case)
the planned maneuver requires the same change in COG as
be before, although the CTW and STW of the own vessel are
different.
Figure 3b. Illustration of the collision avoidance process
when the current direction is in line with the target COG: b -
a 25° turn to the starboard from HDT 010o results in a 035°
new HDT so resulting in smaller SOG of the own vessel, but
accidentally obtains planned DCPA.
3.3 Scenario #3, Example of the Test Executed on the
Simulator
As was explained earlier, different configurations of
ship speeds and currents as well as the relative
arrangement of ships were tested in the simulator.
However, they all referred to rule 15 of COLREG. One
of the variants is described below.
Target vessel is observed at point A, on distance of
8NM. After 6 min the target is at distance of 7NM, so
the relative speed of the target is 10kn. The line
connecting both points (relative course of the target) is
directed to the center of the picture, so exists the risk of
the collision. Speed of the current is 10 knots, direction
090°. If CTW of the own vessel is 340° and STW is
20kn then COG is 010° and SOG 20kn. If the safety
distance (DCPA) is taken as 2NM then own vessel
should turn 30° to the starboard, however changing
heading (not COG!) to the starboard at 30°, the new
heading will be 010
o
and adding up with the vector of
the current, the new COG will change into 040°. In
these circumstances the relative movement of the
target will be differ than expected and DCPA will be
lower than 0.5NM.
Figure 4a. An illustration of the collision avoidance process
when the direction of the current is perpendicular to the own
vessel’s COG: a - planning maneuver COG should be
changed at 30° to the starboard.
352
Figure 4b. An illustration of the collision avoidance process
when the direction of the current is perpendicular to the own
vessel’s COG: b if heading will be changed at 30° to the
starboard then new CTW will be 10
o
and in consequence the
new COG = 040°. If so, then DCPA is about 0.5NM only, not
according to the plan.
4 DISCUSSION
Similar considerations and experiments as one
presented as 3.3 have been carried out for different
proportions of own vessel's speed, target and sea
current. Different mutual positions of the two vessels
were also considered, however, with the assumption
that the target was on the starboard side, at distances
of 8 or 20NM. Scenarios were composed of 8th cardinal
and intercardinal directions of the current in the
relation to the own COG with the combinations of the
own speed 12, 15 and 20kn and encountered vessel of
8, 12 or 24kn.
The tests have been carried out on the navigation
full bridge simulator Navi Trainer 5000 [21] with the
use of warship model (frigate) as an own ship and
coastal cargo ship and fast cargo seagoing vessel
models as encountered vessels. In total 70 scenarios
were tested on the open, calm sea and part of them
consider moderate wind. Depending on the relative
wind direction, the results of some scenarios were
different from others, what was expected.
But in general, they show that low current
velocities, approximately up to 3 knots, do not cause
significant discrepancies in the results of the anti-
collision maneuver, what was predicted. This is
because for vessels moving at speeds in the range of 10-
25 knots, low current does not cause significant
differences between the COG and CTW. Therefore, it
can be assumed that the low current is not significant
for the issues under consideration. However, when the
speed of the current exceeds these values, the use of
information about the COG and SOG for planning the
maneuver and performing it by changing the vessel's
heading (in fact CTW, not COG), causes that, in reality
the CPA may decrease to zero or increase by 100%
compared to the planned value. It depends primarily
on the direction of the current in relation to the COG of
the own vessel and the relative bearing onto target. The
greatest changes occur when the direction of the
current is perpendicular to the own ship's COG, and
the divergence from plan is directly correlated with
current speed.
A synthesis of such considerations is shown in
Figure 5, where the changes between the planned
DCPA value calculated based on the COG/SOG and
implemented by changing the vessel's course are
included.
Figure 5. General tendencies in changes of DCPA as a
function of the differences between the direction of the
current (CD) and COG of the own vessel and the direction to
the target. Changes are presented in % of planned DCPA if
before the maneuver DCPA is zero.
The three presented curves correspond to three
different directions to the target (heading angles) from
the own vessel (the heading angle is marked with a
colored arrow in the color of the appropriate curve).
The values of differences between the planned and
actual value of the CPA distance (DCPA) were
presented as a function of the angle between direction
of the current (DC) and COG of the own vessel and the
relative bearing onto encountered vessel. The graph
shows the changes when the initial DCPA is zero. The
axis showing changes in DCPA ( DCPA) is scaled as a
percentage of planned DCPA. It should be emphasized
that this picture is slightly different if the initial DCPA
value is not zero. Then the changes are smaller by
several percent. The values of these changes depend on
the speed of both vessels and the speed of the current,
as well as the distance between the vessels at the start
of the maneuver and planned DCPA. At this stage of
research, only certain trends can be indicated. The
impact of individual factors and the correlations
between them require further in-depth research.
However, there is a clear tendency to change the actual
DCPA value in relation to the planned one.
This depends primarily on the direction of the
current:
If the current causes an increase in the speed of the
own vessel or when the direction of the current
increases the speed of approach of both vessels, the
actual passing distance decreases as compared to
the planned one,
The closer the target is to the traverse, the greater
these differences are,
If the direction of the current causes the SOG of the
own vessel to decrease, the passing distance does
not change significantly,
If the direction of the current is in the sector 90o
from the direction to the target, the actual passing
distance decreases, while when these directions are
opposite the CPA increases.
Various current speeds have been assumed in the
considerations, but not exceeding 10 knots, as in
relation to standard vessel speeds, the speed of the
current does not exceed the speed of typical vessels. It
has been assumed that the speed of both vessels did not
differ significantly. The differences between the
planned and the actual DCPA differed, sometimes by
353
a factor of two. This confirms the questionable
usefulness of the solution considered and highlights
the need for careful observation of the effects of the
maneuver by the officer of the watch. It can be assumed
that if on a manned vessel the officer of the watch
(OOW) supervises the maneuver, and if it turns out
that there is still a risk of excessive approaching (the
maneuver turned out to be unsuccessful), he will
simply correct the movement of his vessel. However,
in the context of the discussed problem, the risk of
collision with unmanned objects is of particular
importance.
In this context, apart from the previously
considered aspect of using information about the
movement relative to the ground, not the water, there
is also the aspect of possible differences between the
speeds of both vessels. It can be assumed that
unmanned transport vessels MASS will move with
speed similarly to manned vessels and will most likely
be detected at distances like those at which ordinary,
manned ships are detected. However, small,
unmanned vessels (surface drones) engaged in various
research and measurements at sea should be noticed.
As a rule, they are small objects, which cause an
additional risk of being detected at short distances.
But the more important aspect is their low speed, of
the order of a few knots. This means that the speed of
the current radically changes the nature of their
motion. It should be assumed that such objects will also
be equipped with AIS devices, which is one of the
reasons for the increasingly common consideration of
collision avoidance issues based on movement over the
ground, with AIS system. An example of joint use of
AIS and ARPA is given, for example, by [10]., however,
it may be debatable to equip such an object with radar.
Often AIS will be the only source of data for such a
maneuver.
In addition, it should be assumed that there may be
a tendency to limit the range of navigation systems that
these small facilities will be equipped with, so such a
drone may not have a speed log. Consequently, it is
very likely that information about their movement
through the water will be unavailable, and information
about their movement will only result from the use of
the GNSS, which means information about the
movement over the ground. If so, the issue under
consideration becomes a burning one.
The adoption of vessel motion parameters relative
to the ground (COG, SOG) is inconsistent with the IMO
recommendations, however, the negative effects of
such an approach are often insignificant. If external
factors (mainly current and wind) do not significantly
affect the differences between the vector of motion
through the water and over the ground, such an
approach is acceptable. It also turns out to be a
negligible factor when external factors increase the
SOG in comparison to STW.
What is most important, in tidal waters, when the
current speed exceeds 30% of the own SOG the actual
DCPA may vary dramatically from the planned one,
and these differences may exceed 100% of the planned
DCPA.
This phenomenon also occurs on passenger vessels,
ferries and others, which are characterized by a large
windage area, although this issue has not been
described in the article, the possibility of a similar
threat should be considered. This issue requires further
investigation.
A separate aspect of the issue is the possibility of
anti-collision maneuvering of autonomous surface
vehicles, which usually do not have sufficient speeds.
So, in tidal waters the drift can be a significant
impediment.
Presented studies have the character of the
recognition of the problem. Further studies should be
concerned on the case of small surface autonomous
vessels and short distances of detection, as this cause
small value of the distances of closest point of arrival
and consequently biggest threat of collision.
5 CONCLUSIONS
In this paper only one, specific aspect of collision
avoidance problem has been analysed which is the
effect of using the own ship's motion vector over the
ground instead of its motion vector through the water
as recommended by the IMO.
The basic conclusion is that the use of COG and
SOG is permissible at low values of current speed and
when the current direction is consistent with or
opposite to the own COG. This is especially essential,
when encountered vessel is near the traverse of the
own ship.
The problem undoubtedly requires deeper
investigation, especially with the increasing number of
unmanned ships navigating the. The author was
unable to find any publications in global literature
addressing this topic, which appears to be growing in
significance. It would be advisable to conduct research
using the ASV model, if this ship avoided collisions.
Currently, the author does not possess such a
mathematical model which can be used in the
simulator. The correlation between the influence of
many factors on this issue also requires in-depth
research.
Next essential aspect is the lack of the credible
knowledge about the influence of the wind on the
unmanned ship (especially ASV) and the own ship in
the conflict situation. The growing number of small
ASV, some of them partially submerged, opens another
aspect of this issue, as these small units will be detected
at short distances, which may limit the
maneuverability of a seagoing vessels. If we assume
that ASVs will move with speeds not greater than 10
knots, then the speed of the sea current seems be
extremely important factor.
REFERENCES
[1] D. I. Norman, ‘A Short History of Radar at Sea’. [Online].
Available:
https://www.emforensics.com/main/MNRHandbookHis
tory.html
[2] A. Felski, ‘Gyrocompasses - Their Condition and
Direction of Development’, Int. J. Mar. Navig. Saf. Sea
Transp., vol. 2, pp. 5559, 2008, [Online]. Available:
https://www.transnav.eu/Article_Gyrocompasses_-
_Their_Condition_Felski,5,72.html
354
[3] T. Brcko, I. Pavić, J. Mišković, and A. Androjna,
‘Investigating the Human Factor in Maritime Accidents:
A Focus on Compass-Related Incidents’, Trans. Marit.
Sci., vol. 12, no. 2, pp. 112, 2023, doi:
10.7225/toms.v12.n02.w01.
[4] Y. Huang, L. Chen, P. Chen, R. R. Negenborn, and P. H.
A. J. M. van Gelder, ‘Ship collision avoidance methods:
State-of-the-art’, Saf. Sci., vol. 121, no. April 2019, pp. 451
473, 2020, doi: 10.1016/j.ssci.2019.09.018.
[5] A. K. Jadhav, A. R. Pandi, and A. Somayajula, ‘Collision
avoidance for autonomous surface vessels using novel
artificial potential fields’, Ocean Eng., vol. 288, pp. 1–28,
2023, doi: 10.1016/j.oceaneng.2023.116011.
[6] W. Kazimierski and A. Stateczny, ‘Fusion of data from
AIS and tracking radar for the needs of ECDIS’, 2013
Signal Process. Symp. SPS 2013, no. June 2013, 2013, doi:
10.1109/SPS.2013.6623592.
[7] D. Chen, P. Chen, and C. Zhou, ‘Research on AIS and
Radar Information Fusion Method Based on Distributed
Kalman’, in 2019 5th International Conference on
Transportation Information and Safety (ICTIS), 2019, pp.
14821486. doi: 10.1109/ICTIS.2019.8883594.
[8] Y. Zhuo and G. E. Hearn, ‘A ship based intelligent anti-
collision decision-making support system utilizing trial
manoeuvres’, in 2008 Chinese Control and Decision
Conference, 2008, pp. 39823987. doi:
10.1109/CCDC.2008.4598079.
[9] Z. Pietrzykowski, P. Wołejsza, and P. Borkowski,
‘Decision Support in Collision Situations at Sea’, J. Navig.,
vol. 70, no. 3, pp. 447464, 2017, doi:
10.1017/S0373463316000746.
[10] P. Borkowski, Z. Pietrzykowski, and J. Magaj, The
algorithm of determining an anticollision manoeuvre
trajectory based on the interpolation of ship s state
vector , Sensors, vol. 21, no. 16, 2021, doi:
10.3390/s21165332.
[11] Y. He, Y. Jin, L. Huang, Y. Xiong, P. Chen, and J. Mou,
‘Quantitative analysis of COLREG rules and seamanship
for autonomous collision avoidance at open sea’, Ocean
Eng., vol. 140, pp. 281291, 2017, doi:
https://doi.org/10.1016/j.oceaneng.2017.05.029.
[12] W. Naeem, G. W. Irwin, and A. Yang, ‘COLREGs-based
collision avoidance strategies for unmanned surface
vehicles’, Mechatronics, vol. 22, no. 6, pp. 669678, 2012,
doi: https://doi.org/10.1016/j.mechatronics.2011.09.012.
[13] COLREG, ‘Convention on the International Regulations
for Preventing Collisions at Sea, 1972 (COLREGs)’. IMO,
London, 1972. [Online]. Available:
https://www.imo.org/en/About/Conventions/Pages/COL
REG.aspx
[14] Cockcroft A.N. and Lameijer J.N, A Guide to the
Collision Avoidance Rules, Sixth edit. London: Elsevier
Butterworth-Heinemann, Oxford, 2004. [Online].
Available: https://shop.elsevier.com/books/guide-to-the-
collision-avoidance-rules/cockcroft/978-0-08-052192-3
[15] A. Felski, K. Jaskólski, and P. Banys, Comprehensive
assessment of automatic identification system (AIS) data
application to anti-collision manoeuvring, J. Navig., vol.
68, no. 4, 2015, doi: 10.1017/S0373463314000897.
[16] A. Harati-Mokhtari, A. Wall, P. Brooks, and J. Wang,
‘Automatic Identification System (AIS): Data Reliability
and Human Error Implications’, J. Navig., vol. 60, no. 3,
pp. 373389, Sep. 2007, doi: 10.1017/S0373463307004298.
[17] IMO, ‘IMO RESOLUTION MSC.334(90) ADOPTION OF
AMENDMENTS TO THE PERFORMANCE
STANDARDS FOR DEVICES TO MEASURE AND
INDICATE SPEED AND DISTANCE (RESOLUTION
MSC.96(72))’. IMO, London, 2012. [Online]. Available:
https://wwwcdn.imo.org/localresources/en/KnowledgeC
entre/IndexofIMOResolutions/MSCResolutions/MSC.334
(90).pdf
[18] A. Banachowicz and A. Wolski, ‘Determination of Ship
Approach Parameters in the Polar Coordinates System’,
Reports Geod. Geoinformatics, vol. 96, no. 1, pp. 18,
2014, doi: 10.2478/rgg-2014-0001.
[19] A. Bole, A. Wall, and A. Norris, Radar and ARPA
Manual. London: Elsevier Butterworth-Heinemann,
Oxford, 2013. [Online]. Available:
https://shop.elsevier.com/books/radar-and-arpa-
manual/bole/978-0-08-097752-2
[20] M. H. Carpenter, Real Time Method of Radar Plotting,
Subsequent. Maryland: Cornell Maritime Pr/Tidewater
Pub Maryland, 2005. [Online]. Available:
https://schifferbooks.com/products/real-time-method-
radar?srsltid=AfmBOoq97AF6g-
0zeQbENy2fhsMGY2LefPrvBCTTXhQvlg1ZgrSR7TuQ
[21] P. Zwolan and K. Czaplewski, ‘Evaluation of Vessel
Simulation Model Compatibility With Its Equivalent in
Reality’, Annu. Navig., vol. 20, no. 1, pp. 163170, 2014,
doi: 10.2478/aon-2013-0011.