International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 1
Number1
March 2007
101
An Application of ANN to Automatic Ship
Berthing Using Selective Controller
Namkyun Im
Mokpo National Maritime University, Mokpo, South Korea
Lee Seong Keon & Bang Hyung Do
Pusan National University, Pusan, South Korea
ABSTRACT: This paper deals with ANN(Artificial Neural Networks) and its application to automatic ship
berthing. As ship motions are expressed by a multi-term non-linear model, it is very difficult to find optimal
methods for automatic ship berthing. When a ship makes its berthing operation, the ship’s inertia and slow
motion make the ship approach to final berthing point with pre-determined navigation pattern. If the ship is
out of the pre-determined navigation pattern, the berthing usually end in failure. It has been known that the
automatic control for ship’s berthing cannot cope with various berthing situations such as various port shape
and approaching directions. For these reasons, the study on automatic berthing using ANN usually have been
carried out based on one port shape and predetermined approaching direction.
In this paper, new algorithm with ANN controller was suggested to cope with these problems. Under newly
suggested algorithm, the controller can select different weight on the link of neural networks according to
various situations, so the ship can maintain stable berthing operation even in different situations. Numerical
simulations are carried out with this control system to find its improvement.
1 INTRODUCTION
The study on ship automatic berthing control has
been one of the most difficult problems in ship
control fields. When we take a look on a ship’s
berthing operation in certain harbour, the captains
and pilots consider a lot of factors such as ship’s
present speed, heading angle and remain distance to
pier and so on to control ship’s speed and heading
angle with engine revolution and rudder angle.
Therefore ship berthing control is called as one of
MIMO(multi input and multi output) control. In
addition to this, because the ship’s inertial is large
and its motion around pier is slow, the effect of
environments such as wind and current has big
impact on the safety of ship’s operation in a port.
Many researches have been carried out related to
ship automatic berthing problems. H. Yamato[2]
performed successfully the numerical simulation of
automatic ship berthing control using intelligent
control method. He applied feed-back control into
ship automatic berthing problem and found
successful results. Enviromental effects including
wind and current was used in input items to
overcome disturbance problems in this control.
However it was very time-consuming and tedious
works to consider all of wind directions. It was
found that more dedicated control method was
necessary to solve these problems due to
complicated and nonlinear characteristic of ship’s
motion. Another researches [3,4,5] have been
performed to solve ship berthing control using
artificial intelligent theory and its related fields.
When a ship approaches a pier and unberth the
same place, sufficient waters space is needed
because of its own turning characteristic and large
inertial moment. Therefore a ship usually has a
certain berthing pattern, for example, most of ship in
certain pier have similar route with each other or the
ship should have certain heading angle in advance
102
within certain area for successful berthing. When we
take a look on existing researches on ship berthing
control, most of ship approach berthing point with
pre-determined heading angle or route.
In this paper, new approach is suggested to
overcome these problems. Selective controller that is
suggested using ANN controller will guide a ship
toward berthing point even the ship approaches from
multi-direction. Numerical simulation is carried out
to verify its effectiveness.
2 SHIP DYNAMIC MOTION
A type of tanker was adopted as a model, of which
axes coordinates are depicted in Fig. 1. The details
of the tanker and its particulars are presented in
Table 1. Many mathematical models of a ship
motion have been proposed, module type ship
motion (6) is adopted in this paper, because it has
been known that it is suitable to express ship motion.
In order to formulate the equations of ship motion,
two systems are considered where one is the space
axes,
oo
yxO ,−
and the other is the ship body axes,
yxG ,−
, or the moving axes which are fixed in the
center of gravity of the ship.
Fig. 1. Coordinate system for ship dynamics
Table 1. Particulars of ship
Hull
Ship type
Length
Beam
Draft
Cb
Tanker
304 (M)
52.5 (M)
17.4 (M)
0.827
Propeller
Rudder Height
Propeller Diameter
Propeller Pitch
Rudder area
Pitch ration
12.94
8.5
5.16
98.0
1.709
The maneuvering of a ship is usually described in
the form of the modular mathematical model, in
which the total hydrodynamic forces and angular
moment are split into separate parts. Consequently,
the ship’s dynamical behavior can be described by
the equations of motions for three degrees of
freedom as follows:
WRPHyx
XXXXvrmmumm +++=+−+ )()(
(1)
WRHxy
YYYurmmvmm ++=+++ )()(
WRHZZZZ
NNNrJI ++=+
)(
where;
HHH
NYX ,,
: Hydrodynamic forces and angular
moment acting on a hull
RRR
NYX ,,
: Hydrodynamic forces and angular
moment due to the rudder
P
X
: Hydrodynamic forces due to the propeller
WWW
NYX ,,
: Hydrodynamic forces due to wind
3 SHIP AUTOMATIC BERTHING CONTROL
SYSTEM
3.1 Artificial Neural Networks design
When a ship is under berthing operation, various
factors are considered such as speed and heading and
so on for successful ship’s berhing. Fig. 2 shows
brief ship berthing illumination and items to be
considered such as speed and distance. As shown in
this figure navigators considers ship’s speed,
heading angle and remain distance to berthing point
and so on as input factor to control ship until a ship
arrive final point.
Input layer
Hidden layer Output layer
ξ
η
v
r
ψ
1
d
u
2
d
δ
n
jk
W
,
k
I
ij
W
,
i
O
j
a
Fig. 3. Construction of berthing controller in ANN
Total 8 of input items are considered as follows;
lateral and longitudinal distance (
ηξ
,
),Lateral and
longitudinal speed (), turn rate (r), heading angle
(
ψ
), distance to goal point (
2
d
) and vertical
103
distance with approach line to berthing point(
1
d
).
Two item of output are engine revolution per second
(n ) and rudder angle(
δ
). Fig. 3 shows final diagram
of neural networks used in this research. Here
ijjk
WW
,,
,
indicate links coefficients,
ijk
OaI ,,
means input layer, hidden layer and output layrer
respectively.
The number of neuron in hidden layer was set to
15 that is bigger more than that of input layer. The
value of input, output and weight are normalized
between 0 and 1. The propagation method is adopted
as learning way. Sigmoid nonlinear function also
used in neural networks. Matlab 6.0 was used as the
programming language for numerical simulation.
3.2 Structure of ship automatic berthing system
When a ship approaches to pier and is under berthing
operation, the ship usually maintains certain heading
angle toward final berthing point. It is similar with
air plain that is about to take on the ground. The
phnomiem is maintained for the safety of ship and
air plain. The mass of ship is larger than other
vehicles in a land and it speed is greatly slow under
its berthing operation. It requires spare water space
to make it turns and rudder effect extremely down
when its speed is under certain limit line.
Therefore when we take a look at previous
researches on neural networks controller, most of
them show such as study that a ship approaches the
final berthing point from very limited direction. In
other words, when we take an example like Fig. 4,
the ship approaches the berthing point through
“Area-1” in many study case. This is the reason that
it is physically impossible to make a ship turn or to
maintain the ship with desired heading angle with
slow speed within certain distance to berthing point.
B
A
Berth Point
Area-1
Area-2Area-3
Fig. 4. Berthing concept
Be r t hi ng
condt i ons?
Yes
No
Suggest
re-tri al
Appr oachi ng
Saf e Ar ea?
Yes
Cont r ol l er f or
Ar ea 1, 2
No
Be r t hi ng
Oper at i on
Cont r ol l er f or
Ar ea 3
Tur ni ng Shi p
Fig. 5. General flow of berthing control
In this research, new approach is suggested to
overcome these problems. New system is suggested
to guide a ship to final berthing point with selective
controller even the ship come from multi-direction.
The Fig. 5 shows its brief concept.
When a ship approaches goal point, the system
reasons whether it is possible to make a berthing or
not. At this moment ship’s present conditions such
as present location, speed and heading angle are
used. If the system estimate berthing is impossible
under present conditions, it suggests the ship make
turn and approaches new areas.
If the present conditions are clear, assigned
controller at each area is used to make the ship berth.
For example, when a ship approaches to Area-3, the
possibility of berthing is almost impossible, because
the remained distance to final point is too short and
the ship have to make turning under slow speed. In
this case, the system guides the ship make a turn
toward area 3, area-B, Area-A and Area-2. If a ship
enters into Area-2 that is safe zone for berthing,
assigned controller will guide the ship to final
berthing point.
In other hands, a ship approached from Area-1 or
Area 2 originally will be guided toward berthing
point by assigned controller in each area because
these areas are safe zone for berthing.
104
4 NUMERICAL SIMULATION
4.1 Automatic control at each zone
The figure 6 shows 4 cases of numerical simulations
when a ship advances to berthing point through
“Area-3”. As shown in this figure, the ship is
controlled from Area-3 to Area-B by assigned
controller of Area-3. Additional 4 cases of numerical
simulations were carried out to verify the controller
from Area-B to Area-A, the results are shown in fig.
7. It is found that all cases are successful to move a
ship between two areas. It is peculiar that even all
ship departs from different point they arrived at
almost same point in Area-A.
Fig. 8 shows numerical simulations case when a
ship move from Area-A to Area-2 that is safe zone
for berthing. Assigned controller to Area-A was used
to guide the ship and it is found that all cases made
successful shifting of ships. When the ship is con-
trolled safely from other zone to safe zone, Area-2,
successful berthing control could be expected Fig 9
and 10 shows similar simulations results when ships
start from safe zone, Area-1 or Area-2 toward final
berthing point. As shown in figure 9, all ships
maintain similar heading angle around final berthing
point with slow speed, even they departed different
point with different heading angle. In addition to
this, other cases when ships approach to Area-1 are
shown in Fig. 10.
Generally speaking when a ship is under berthing
operation the ship approach the berthing point from
side direction as shown in the figure and maintains
certain heading angle toward the point with decrease
of speed, it is typical pattern in ship berthing
process. This figure shows even ships approach from
different direction with different heading angle, their
berthing operation finished in successful pattern.
-8 -6 -4 -2 0 2 4 6 8 10
0
5
10
15
Goal Point
Fig. 6. Simulation Result in Area-3
-8
-6
-4
-2
0
2
4
6
8
10
0
5
10
15
Goal Point
Fig. 7. Simulation Result in Area-B
-5
0
5
10
0
2
4
6
8
10
12
14
16
Goal Point
Fig. 8. Simulation Result in Area-A
0 2 4 6 8 10
-1
0
1
2
3
4
5
6
7
8
9
Goal Point
Area-3
Area-2
Area-1
Fig. 9. Simulation Result in Area-2
105
-2 0 2 4 6 8
-2
-1
0
1
2
3
4
5
6
7
8
9
Goal Point
Area-2
Area-1
Fig. 10. Simulation Result in Area-1
4.2 Automatic control for all of zone
Numerical simulations also were carried out when a
ship approach from non-safe zone. The many
purpose is to confirm that the suggested selective
controller can guide the ship from dangerous zone to
safe zone, Area-2 and Area-1. The results are shown
in Fig. 11.
-8
-6
-4
-2
0
2
4
6
8
10
0
2
4
6
8
10
12
14
16
Goal Point
Fig. 11. Simulation Result of all- direction- approach case
As shown in the figure, total 4 cases of simulation
scenario were planned. All ship start from Area-3
that is recognized as dangerous zone for safe
berthing. In real world, a ship usually does not berth
in this pattern, because the ship require spare water
area to make a turn and slow down its speed around
final berthing point. Therefore in this scenario a ship
need to make a turn and approach to safe zone such
as Area-2 or Area-1. All simulation results show that
ships approach to safe zone of Area-2 via midterm
areas, Area-A and Area-B. It means that all ships are
controlled with safety from dangerous zone to safe
zone even they started at undesirable zone for safe
berthing.
5 CONCULSIONS
This paper deals with automatic system for ship
berthing. The controller of artificial neural networks
existing researches on ship berthing control, most of
ship approach berthing point with pre-determined
heading angle or route. In this paper, new approach
is suggested to overcome these problems. Numerical
simulations were carried out to verify its
effectiveness. Selective controllers are designed at
each area and guided a ship toward berthing point
even the ship approaches from multi-direction. It
was found that new system makes a ship turn and
guides it toward safe zone for berthing.
REFERENCES
Koyama T. and Jin Y., “A systematic study on automatic
berthing control (1st report),” Journal of the Society of
Naval Architects of Japan, vol. 162, pp. 201, 1987.
Yamato H., “Automatic Berthing by the Neural controller,”
Proc. Of Ninth Ship control Systems Symposuium, vol. 3,
pp. 183-201, May, 1990.
Hasegawa K. and Kitera K., “Mathematical Model of
Manoeuverability at Low Advance Speed and its Appli-
cation to Berthing Control,” Iproc. Of The 2
nd
Japan-Korea
Joint Workship of ship and Marine Hydrodynamics, Osaka,
Japan, 1993.
Namkyun IM. and Hasegawa K., “A study on Automatic Ship
Berthing Using Parallel Neural Controller(2
nd
Report),” The
Journal of Kansai Societiy of naval Architects of Japan, vol.
237, pp. 127-132, Mar. 2002.
Choi Yong-woon et al, “Real-time Detection Techique of the
Target in a berth for automatic ship berthing,” Journal of
Control, Automation and System Engineering Vol, 12 No. 5
May, 2006, pp. 431-437.
MMG, “MMG Report 1-4”, Bulletin of the Society of Naval
Acchitects of Japan, No. 575, 1977.
