Content uploaded by Holger Korte
Author content
All content in this area was uploaded by Holger Korte on Sep 20, 2016
Content may be subject to copyright.
From Nautical Path Planning in ECDIS to Their Realisation
Applied for Fully Actuated Ships.
Oliver Köckritz*, Martin Kurowski***,
Dennis Grunert*, Holger Korte*, Axel Hahn **
*Dept. of Maritime Studies, Jade University of Applied Sciences, Wilhelmshaven/Oldenburg/Elsfleth
Germany (e-mail: oliver.koeckritz@jade-hs.de).
** Department for Computer Science, Carl von Ossietzky University Oldenburg,
Germany (e-mail: hahn@wi-ol.de)
*** Institute of Automation, University of Rostock,
Germany (e-mail: martin.kurowski@uni-rostock.de)
Abstract: New Controllers for ships developed for their modern steering devices need new human
interaction within the controller interface to make these controllers applicable. To increase their
acceptance and usability especially for high precise operations like berthing, a modified path planning
approach shall be introduced which originally bases on the established path planning strategy in
electronic sea charts. The represented planning procedure dispense on parameter based modelling due to
their obscurity while developing electronic tools and their strong variance of cargo ships. Furthermore,
this paper describes the underlying data and the transfer of results from the nautical path planning to a
modern controller device on the example of the MIMO controller system AdaNav, which is applicable
for the complete speed range. The AdaNav controller enables DP equivalent functions for ships being
able to traverse. With that approach the authors want to guarantee full responsibility of the nautical staff,
the support of mates mental model connected to path planning and manoeuvring behaviour and feasibility
to the executing controller.
Keywords: manoeuvre planning, ship guidance system, track control, automatic berthing, marine control,
ship navigation, human centred design.
1. INTRODUCTION
In the last decades maritime ship-control systems have
undergone a remarkable progress from the Nomoto course
control (1957) to 3 (or more) degrees of freedom (DoF)
controlling for fully actuated ships, e.g. with vertical motion
damper in Auestad et.al. (2014) or with active roll reduction
in Koschorrek et.al. (2015). Assuming that present controllers
for marine vehicles are able to realise precise motion steering
of a ship in 3DoF, there are still difficulties to install an easy
useful tool with motion control qualities on a wide range of
maritime vehicles.
On the one hand, the system behaviour-dependency between
the controller characteristics relating to the underlying
mathematical model and the real ship behaviour could be a
reason for such difficulty in their appliance. On the other
hand, insufficient usability and inconsistent acceptance by the
nautical personal on a ship could be a difficulty for its
success. Caused by this leak, controllers are only still used in
open sea situations or for special purposes like Dynamic
Positioning (DP). For manoeuvring and other daily uses, it
has to be taken into account that the navigator is still
responsible for incidents in a moralistic and substantive point
of view. In these cases, the navigator only uses an assistant
system if he feels comfortable with the system behaviour.
This factor is strongly dependent on the mental model of the
navigator. The latter again with learned knowledge about ship
manoeuvrable boundaries while planning and controlling
manoeuvres.
To close the gap between currently grouped use cases by
minimising the dividing factors, this paper describes a track
planning method, which is related to the manual path
planning on a paper sea chart by the use of manual tool
equivalents like triangle and divider to create a kinematic
sequence, which is compatible with modern controllers able
to support DP functionality. Based on that, this paper shows a
new way to generate the relevant process values for the
controller in the chosen setting.
2. STATE OF THE ART
Current commercial planning methods use standard
waypoints and its parameters to describe the path. Thereby a
number of waypoints including a set of attributes are
typically connected with straight lines. To modular course
changes, circular arc elements are automatically integrated in
dependence of the waypoint (WP) attributes. These kinds of
track planning method are in conformance with the current
European standard EN 62065:2002 (CENELEC, 2002). To
generate the turning circle, the smallest radius is defined for
the ship. This definition doesn’t refer to possible
manoeuvring circles in low speed and other behavioural
conditions. While planning a track for the ship, it is not
possible to plan adequate tracks for all manoeuvres or
situations without the use of an entire parameter set. Another
problem can occur in narrow water where waypoints might
have to be set in restricted areas due to requirements for the
desired circle to be inserted. The route cannot be activated by
the operator in commercial ECDIS for track control
application. Besides these problems and some other
weaknesses, commercial track planning solutions don't
support traverse motion and independent heading settings,
which are needed for realistic berthing manoeuvres.
Another commercial application field takes a different
perspective on planning a track of a ship, which is
implemented in DP-systems. This planning still bases on
fixed waypoints and inserted circular arc elements. But with
the addition that the set values are more differentiated:
Instead of setting course and speed, the course is separated in
speed over ground (SoG) and heading. Due to the separation,
the inserted circles refer to the heading changes at a
waypoint. The radius of each circle depends on the
parameters of each waypoint. In K-Pos DP-System
(Kongsberg Marine AS, 2007), these parameters can be set
manually or automatically. The automatic settings are
calculated on the given mathematical model of the specific
ship for each DP-system. The requirement of a concrete
mathematical model is one reason for high development costs
for each DP-system, which results in less numbers of vessels
equipped with DP-systems.
Separated from the commercial shipping, research of path
planning for unmanned vehicles and for groups of vehicles
took place. These approaches for path planning are hardly
connected to their research topic. For cluttered environments,
the approach by Pedersen and Fossen (2012) is based on
potential theory. Zizzari et al. (2010) use incorporating
constraints to plan conflict free paths for multiple marine
vehicles.
New research work is done related to the “line-of-sight”
(LoS) guidance. Lekkas and Fossen (2014) try to minimise
the accumulating cross-track-error caused by a liberal
parameterisation used by the automatic path generating.
Lekkas and Fossen (2013) discuss better approximation to a
given path and extending modelled motion to 6DoF.
In difference to the commercial field, the vehicles in the
mentioned research field are much smaller and better
actuated. The resulting high dynamic characteristics of these
vehicles lead to a much better controllability and therefore to
a wide range of possible path plannings. In all proceeds of the
related research field no nautical officer is responsible for the
resulting planned paths and furthermore developers using
their own mathematical approximations separated within
their components - the controller and the planning
procedures.
The research for precise path planning for commercial ships
is still in progress by using the developed prediction tool of
Benedict et al. (2010). It is based on a simulation of the ship
motion using nonlinear differential equations and its
parameters. By connecting simulated manoeuvres, it would
be possible to generate a completely trackable path based on
comprehensible manoeuvres. For this approach, a sufficient
mathematical model of the used ship is necessary as well.
The enhancement of manoeuvre planning with realism could
increase the acceptances, but the procedure still doesn’t
reflect the navigator’s mental model of planning a track. To
integrate the navigator’s mind it is necessary to use familiar
leg types like straight-lines and circles for instantiations and
accelerations and distances for ship behaviour.
To accomplish our research goal of a user centred path
planning, the envisaged “controller realised planning
procedure” should cover the following conditions:
- Correspond to the navigator’s mental model for path
planning.
- Correspond to the navigator’s mental model in
manoeuvring behaviours.
- The user interface should be in conformance to the given
interface in commercial applications, like the ECDIS
performance standard CENELEC (2002).
- As much as necessary, as less as possible use of
mathematical models.
- Path representation must be confirmed with the usage of a
separated universal controller.
- The path segments should be inherently consistent
arrangeable.
- All needed manoeuvres from berth to berth should be
realisable.
3. SYSTEM SETTING, GOAL AND BACKROUND
The mentioned problem in the chosen setting includes two fix
“interfaces”. The first one comes from the human navigator
and the second one from the AdaNav-controller developed as
triple cascaded MIMO-track controller (Multi-Input-Multi-
Output), which was proposed by Korte et al. (2009). This
controller is designed for fully actuated ships and is able to
correct path deviations by either using heading corrections or
by changing the traverse speed.
Fig. 1: The needed cases of human centred nautical path
planning and realisation with the given MIMO-Controller, in
a collaboration Diagram. The green boxes show the existing
system parts and the blue boxes are met by this research.
As shown in Fig. 1 the navigator as human being plans the
path with accelerations and movement limits in his mind by
setting the parameters graphically and numerically in the user
interface. The interface for the navigator is integrated in a
standard user interface on bridges like the Electronic Chart
Display and Information System (ECDIS). By that way the
now electronically instantiated track is stored in a so-called
kinematic sequence (introduced in Korte, 2014), including
parameters such as accelerations, speeds, lengths and times
between waypoints. From this point of view a waypoint
describes the joint between circles or straight lines from zero
to s (as length) in a continuously mathematic description. The
kinematic sequence is then transmitted to the controller
environment for generating the set values for the MIMO-
controller with DP functionalities. Hereby the mathematical
interpretation of the kinematic sequence in both
implementations in planning and controlling must be
identically to get the right result out of the overall system.
To fill the gaps related to the usability and functionality
within commercial path planning systems as mentioned in
Section 2, the approach for the new path planning system in
this paper is to realise low and high speed track control
including the same existing DP functionalities for low speed
and unify all together in one user interface for seamless use.
The usability should cover a common interaction and use for
a wide range of vessels with different manoeuvring
characteristics.
The real differences shall be compensated by the MIMO-
speed controller. Furthermore, the nautical staff should be
able to plan feasible manoeuvres in narrow water with only
small manoeuvring space and without detailed knowledge
about the mathematical model of the vehicle. The expertise
should base on constraints of possible accelerations.
Fig. 2: The angles and the velocities in the body fixed and
earth fixed frame.
A further assumption is that Kirchhoff Equations of motion
are valid for a ship moving in water; see Scharnow (1987)
and Fossen (1994). Track guidance systems for maritime
surface vehicles use the model background of a slim rigid
body. To realise the 2D path following task the movement of
ships is restricted to the nearly undisturbed water surface.
With surge velocity
u
, sway velocity
v
and yaw rate
r
, a
3DoF system is defined. Figure 2 shows angular and vector
relations of the motion for a surface vehicle in the earth-
fixed, as well as in the body-fixed coordinate system.
4. NAUTIC PATH PLANNIG FOR KINEMATIC
SEQUENCES WITH A NEW SET OF ATTRIBUTES
Relating to the controller design the additional simplification
in case of the traverse motion while course alteration for
3DoF in Eq. (1) is also acceptable for path planning
procedure. In that way, the speed components can be
addressed independently. Lines, circles or other curvilinear
trajectories can be executed by linking the different speed
gradients together.
(1)
In order to receive manageable path segments, it is intended
to restrict the path sections to straight lines and circular arcs.
This restriction is necessary while creating the reference path,
because then the path can be handled and checked by the
ECDIS. Additionally, sharp bends are allowed, which can be
realised by the linear and angular velocities
. In that
case, natural deviations from the reference path are treated as
disturbances, which are compensated by the control system
and should be considered as path safety areas during the
checking process in the ECDIS.
The calculation of the particular path sections is done based
on the kinematic movement of a rigid body on the water
surface. The transitions of the different segments are
calculated involving the manoeuvring characteristics of the
vessel. These characteristics are only operating point based
limitations of maximal positive or negative accelerations.
Figure 3 shows the simple example of planning a transversal
speed change manoeuvre. WP1, WP2 and WP3 are planned by
the nautical staff as geo-referenced waypoints with heading
, completed with the set of commanded
velocities
and additionally acceleration
attributes
. These acceleration attributes have to
stay below the physical limits
Alim,i (
op)
which depend on
loading conditions and operating point limitations, see Eq.
(2). Alternatively, these accelerations can be defined as
constants between two waypoints and calculated from the
commanded velocities and time differences, see Eq. (3), (4).
Fig. 3: Planning example and process differences for a
transversal manoeuvre section.
Being within these operational limitations
lim (
op), it must
be possible for the controller to realise the manoeuvre with
corrections of setting values around the operation point,
including possible forces and moments related to the
acceleration limits in Eq. (2).
(2)
(3)
(4)
During the entire planning process, the different trajectories
are connected to kinematic sequences. Positions, velocities
and time values are strictly joined together. Partially
overlapping periods, which are in conflict with the kinematic
sequences, are already excluded in the planning phase.
Thereby, the manoeuvre strategy is affected by the
configuration of the propulsion, the steering equipment and
their dynamics. It is evident that a ship with pod drives and
transverse thrusters, which is fully actuated in low speed
mode, has excellent manoeuvring characteristics and is able
to turn on the position or move sideways. In contradiction, a
vessel with conventional propulsion and rudder equipment is
not fully actuated - the lateral speed arises indirectly, but has
to be taken into account. Thus, the applied manoeuvre
strategy remains in responsibility of the nautical staff.
Finally, the properties of the additional manoeuvre waypoints
WP11 and WP21, in case of a changing transversal velocity,
can be calculated by the planning method with Eq. (5), (6), in
accordance with the physical limits
Alim,i (
op)
.
(5)
(6)
In comparison to the traverse motion, Fig. 4 shows the
planning sequence of a free running turning circle. The
so-called wheel-over-point (WP11 in Fig. 4) describes the
point of overtaking the steering device, which starts the
accelerations at the half duration period before reaching the
circular arc. The end of acceleration point (not shown in Fig.
4) has the time delay of the half duration, but without any
activity. This results by the agreement of Eq. (1). Before
finishing turning on WP3, the ship starts on WP21 to decrease
the turning rate.
Fig. 4: Planning example and process differences for a free
running turning circle section.
A planning strategy to calculate the starting or ending point
of a manoeuvre uses the final or starting states with constant
accelerations over time. That strategy is necessary for
manoeuvres like berthing to or sailing from zero speed and
can be used for approaching turning circles to start and end
traversal velocity on a given position. Because modelled by
the same waypoint attributes, that strategy does not need
more state functions. Due to the continuity of the kinematic
sequence, it is easily possible to connect parts of different
kinematic sequences by bringing the endings into the same
state. By changing attributes of waypoints graphically or
numerically, the continuity is also guaranteed by
recalculating all effected waypoints.
5. GENERATING THE RIGHT COMMAND VALUES
OUT OF THE PATH
The result of path planning is an enhanced waypoint list as
kinematic sequence covering the
Fset,n
-value of (7). It is
transmitted to the controller device and stored in its own
waypoint list, shown in Fig. 5. Using that internal waypoint
list the controller module state function creator selects an
actual vector of reference functions of the current path
segment depending on the route length
s
, see Eq. (8).
(7)
(8)
The result can be transformed completely to the running time
t
within a completely position controlled process, including
longitudinal compensation of length differences
.
Fig. 5: Schematic diagram of the developed manoeuvre
planning and track guidance system.
The deviations of the planned process and the real process
lead to asynchronity between the route advancement s and the
related
t
. To keep the relation of
s
synchronised to
t
, the
running time
t
must be recalculated from the relation of the
real ship position and the resulting
s
to a new running time,
now called statetime. The statetime is used by assuming the
actual segment of the path, as shown in Fig. 6.
Fig. 6: General procedure of the new state function creator.
Furthermore, the statetime has no chance to get lost from the
real process by being updated in every loop. By the use of the
resulting actual set functions, the called module command
value generator will be able to compute the real time related
command values for the different subordinated control
cascades, e.g. thrust control
[X,Y,N]set
, velocity control
[u,v,r]set
, heading control
set
, and track control
,
additionally time-dependent control deviations with respect
to the actual geographic position (
)
act
, heading
act
, and
velocities
[u,v,r]act
which are necessary for the computation
of controller action.
The black box MIMO controlled marine vehicle in the lower
part of Fig. 5 represents the track controlled vessel using the
3DoF speed-adaptive ship model, as well as the developed
modular, adaptive MIMO track control device.
Both defined geometric figures for the planned vehicles path
allow a simple computation of the route length s and cross-
track error
xte
. For the arc segment, the solution is given by
the triangular relation of the actual waypoint, the rotation
centre point and the current position, whereby the leg
distance
is proportional to the angle
on the centre
point and
xte
is the difference of its leg lengths. In case of a
straight leg, both values can be calculated by the solution of
the point – a directed shape of a straight line function added
with their normal vector to the present position. Having
calculated the route length
s
from the actual position
measurement (
)
act
, all other control deviations can be
calculated from the vector of set value functions, see Eq. (8),
and the process value measurements.
Fig. 7: Detail view on the route plan of an approaching and
mooring scenario in Rostock harbour. The mooring at place
64 is based on recorded data of a manual berthing manoeuvre
of the ferry M/V Mecklenburg-Vorpommern.
(Screenshots of ECDIS with integrated new planning tool)
6. RESULTS
The implementation of the path planning and the integration
of the MIMO-controller were completed during the project
“Integrative Manoeuvre Realisation system for Ships”
(IMaReS). For ECDIS proposals, the SevenCs ECDIS-Kernel
5.14.6.0 is used. The new user interface for path planning still
has to be fully evaluated, but it is already possible to plan
easy and intuitive correct manoeuvres like berth-to-berth or
arrival and berth. First evaluations show that nautical staff is
able to use the interface intuitively after a five minutes long
introduction and by this in a much faster way than with
common DP interfaces.
Figure 7 shows screenshots of an approach and berth scenario
planned with the developed tool. In Fig. 7 (right) the
additionally inserted WPs can be seen as small circles. The
bigger circles represent the waypoints set by the user. The
different angle of the legs connected to WP19 in Fig. 7 (right)
is caused by a traversal manoeuvre.
Fig. 8: Simulation results of an automatic to sail manoeuvre
in the port of Rostock in different scales. (Screenshots of
ECDIS with integrated new monitoring tool)
For the simulation trials, a test bed consisting of a real-time
application, running on a standard industrial-PC and
programmed with MATLAB/Simulink and xPC-Target is
used. Furthermore, the monitoring data is send back to the
ECDIS for visualising the track and process. Figure 8 shows
the simulation result of the MIMO controlled vessel m/f
“Mecklenburg-Vorpommern”, stimulated by the planned
route by means of the developed planning tool. Connected by
straight lines, the bubbles in red depict the planned track and
the small dotted blue line shows the system answer of the
controlled vessel in the pictures. The left picture shows an
overview of the monitored simulation. The centred picture
and its scale suggest an excellent performance of the used
planning and control structure. As seen in a closer look to the
right picture in Fig. 8, a small deviation exists between
commanded track and the vessel response, which can be
caused by insufficient ship and controller parameters, e.g.
accelerating effects.
7. FURTHER WORK
The complete set of variables included in Fset,n can be used to
include nonlinearity inside a segment by changing the used
dependency between the velocities and its related
accelerations. The resulting parabola velocities inside a
segment could be used to imitate a common manoeuvring
behaviour of nautical staff in berth situations. The mate
would only have to increase the relevant acceleration
parameter of the related waypoint. The result is to keep the
nautical officers in their mental safety zone while automated
ship manoeuvring and therefore gains a good acceptance of
the complete planning and control system.
To prove and improve the user interface, sustainable empiric
studies have to be done with the new interface. For this
reason, further work has already been started on the
integration of the path planning tool into a nautical
environment in order to use it permanently and for the
empiric research.
To prove the complete functionality of the system in the real
envisaged setting, also tests have to be done with ships in a
maritime test environment. For instance the research harbour
of Rostock could host this request, because of the installed
precise local referenced positioning system, which is needed
for precise and automated manoeuvring as shown in Fig. 7.
To improve the applicability of the used MIMO-Controller
for different environments, ships and its conditions, research
on the subsystems and architecture already has been started to
get the achieved widely applicable.
8. CONCLUSION
The paper presented the implementation of a new full-state
kinematic sequence based track planning for ships, necessary
for berth-to-berth navigation. The simulation of typical to sail
manoeuvres in the port of Rostock, realised and observed
with the implemented tools, shows well comparable results to
practical manoeuvres and the potential for automatic ship
guidance.
Unlike to model based planning algorithms only boundaries
of motions have to be considered by the nautical stuff.
Nautical student research has shown that the aim to match the
mental models of nautical stuff is in focus. Implementing the
planning and monitoring tool in an ECDIS give the best
opportunity to be in conformance with ECDIS performance
standards. To simplify the mathematical background to
geometry math for calculating the parameters of the
Waypoints makes it easy to arrange the path segments
inherently consistent.
But accurate track guidance is not possible without equally
accurate and stable position, velocity and heading
information. Therefore, further investigations are necessary
for the integrity of measurements in such safety-relevant
applications.
ACKNOWLEDGEMENTS
The authors would like to thank the Federal Ministry of
Education and Research for supporting the project IMaReS
(Reg. no. 17014 A 11) and the SevenCs Company for support
in software licences and practical backing.
REFERENCES
Auestad, Ø. F., Gravdahl, J. T., Sørensen, A. J., Espeland, T.
H. (2014). Motion Compensation System for a free
floating Surface Effect Ship. 19th IFAC World Congress,
24.-29. Aug. 2014, Cape Town, South Africa, 6 pages.
Benedict, K., Gluch, M., Kirchhoff, M., Fischer, S., Klaes, S.,
and Baldauf, M. (2010). Application of Fast Time
Simulation Technologies for Enhanced Ship
Manoeuvring Operation. In Proceedings of the 8th IFAC
Conference on Control Applications for Marine Systems,
79-84, Rostock, Germany.
CENELEC (2002). Track control systems. European
Standard EN 62065:2002, European Committee for
Electrotechnical Standardization
Fossen, T. I. (1994): Guidance and Control of Ocean
Vehicles. Verlag John Wiley & Sons, Chichester / New
York, ISBN 0-471-94113-1.
Kongsberg Marine AS (2007) K-Pos Dynamic Positioning
System, operator manual, rel. 7.0, Kongsberg, Norway,
Oct. 2007.
Korte, H., Kurowski, M., Baldauf, M., and Lampe, B. (2009).
AdaNav - A Modular Control and Prototyping Concept
for Vessels with variable Gear Configurations. In
Proceedings of the 8th IFAC Conference on
Manoeuvring and Control of Marine Craft, pp. 91-96,
Guaruja, Brasilia.
Korte, H., Köckritz, O., Kurowski, M. (2014). Kinematic
sequence based track planning for ships and marine
vehicles. International Conference on Information
Science, Electronics and Electrical Engineering (ISEEE),
Sapporo, Japan, pp. 545-550, ISBN 978-1-4799-3196-5.
Koschorrek, P., Siebert, C., Haghani, A., Jeinsch, T. (2015).
Dynamic Positioning with Active Roll Reduction using
Voith Schneider Propeller. In Proceedings of the 10th
IFAC Conference on Manoeuvering and Control of
Marine Craft, pp. 178-183, Copenhagen (DK).
Lekkas, A. M. and T. I. Fossen (2013). Line-of-Sight
Guidance for Path Following of Marine Vehicles.
Chapter 5, In "Advanced in Marine Robotics", LAP
LAMBERT Academic Publishing (O. Gal, Ed.), 2013,
pp. 63-92. ISBN 978-3-659-41689-7.
Lekkas, A. M. and T. I. Fossen (2014). Integral LOS Path
Following for Curved Paths Based on a Monotone Cubic
Hermite Spline Parametrization. IEEE Transactions on
Control Systems Technology, Vol. 22, No. 6, 2014, pp.
2287-2301.
Nomoto, K., Taguchi, T., Honda, K., and Hirano, S. (1957):
On the steering qualities of ships, Int. Shipbuilding
Prog., vol. 4, no. 35, pp. 354-370.
Pedersen, M. and Fossen, T. (2012). Marine Vessel Path
Planning and Guidance Using Potential Flow. In
Proceedings of the 9th IFAC Conference on
Manoeuvring and Control of Marine Craft. Arenzano,
Italy.
Scharnow, U. (1987): Schiff und Manöver. Seemannschaft
Bd.3, Transpress Verlag, Berlin, 3. ed., ISBN 3-344-
00151-5.
Seven C’s (2007) EC 2007 Kernel, Version 5.14.6.0,
Hamburg, 2007.
Zizzari, A., Pedone, P., and Indiveri, G. (2010). A Path
Following Controller for the Dynamic Model of a
Marine Surface Vessel. In Proceedings of the 7th IFAC
Symposium on Intelligent Autonomous Vehicles, 491-
496, Lecce, Italy.
