Return to Session Directory

Authors Name

Name of the Paper

Session

DYNAMIC POSITIONING CONFERENCE

October 12-13, 2010

NEW APPLICATIONS SESSION

Improved usability in DP tracking operations

By Ivar-Andre Flakstad Ihle

Rolls-Royce Marine AS
lesund/Oslo, Norway

Ivar Ihle

New Applications

Improved DP tracking usability

Abstract
This paper presents examples of improved usability in a low-speed waypoint tracking system.
The improvements have partly been rendered possible by a path construction scheme where
the type of path and speed profile provide complete path information. Combined with
functionality, this ease the tracking operation and reduce operators workload. Path
construction, path properties and situations with several, maybe conflicting, assignments are
presented and discussed. Usability examples of both functionality and the graphical user
interface characteristic are provided.

Introduction
Waypoint tracking is a natural extension of the position and heading control modes of a
dynamic positioning system. It controls the vessel along a track of predefined waypoints and
the operator can usually choose between different heading and speed control modes. In lowspeed mode (speeds up to 3-4 knots) waypoint tracking is an important component of
advanced vessel operations such as cable and pipe laying, dredging, trenching, and surveying
[2].
In the context of this paper, usability is part tracking functionality and part user interface
characteristic. It has been a goal throughout the design process to focus on the user and much
thought has been put into how different tools, such as path analysis and construction, can ease
the operators workload.
To achieve waypoint tracking while being able to incorporate vessel speed and acceleration
and other dynamic constraints, analyze path properties in a straightforward manner, and
facilitate an advanced interface implementation, new functionality has been developed:
methods for path construction, speed profile generation, and tracking analysis tools. This
paper will show usability examples rendered possible from the underlying functionality, GUI
examples, and other elements that simplify often-used operations.
The background of this paper is the design, development and implementation of a low-speed
waypoint tracking system. More specifically, we consider some of the important aspects of
the reference model and the user interface. Most of the material is also applicable for highspeed waypoint tracking systems. The difference is mainly the number of controllable
degrees-of-freedom (DOFs). In the underactuated case, it is impossible to control all DOFs
directly. A vessel may be underactuated because of insufficient thruster arrangement or
speeds that render the tunnel thrusters ineffective. For instance, a platform supply vessel at
zero or low speed is typically overactuated and controllable in surge, sway and yaw.
However, at high speeds the sway motion can only be controlled indirectly so the system is
said to be underactuated.

From points to paths

Waypoint tracking is an essential part of the reference module in the DP control system. It
must provide continuous reference signals for position and velocity to maneuver the vessel
from waypoint to waypoint.
A waypoint consists of at least a position (e.g., latitude and longitude, UTM, etc.), and
perhaps a turning radius, speed and heading setpoint. For a DP tracking system, the waypoints
usually comes from an external source.

DP Conference

Houston

October 12-13, 2010

Page 2

Ivar Ihle

New Applications

Improved DP tracking usability

The radius of a waypoints acceptance circle, where the vessel starts turning towards the next
waypoint is related to the turning radius

r turning r acceptance tan

where is the angle between previous and next waypoint.

Waypoint with acceptance and turning radius

The goal for the DP control system is to convert the waypoints into a path that can be
followed by the vessel. The classical approach is to connect the waypoints using straight lines
and inscribed circles determined by the turning radius. There exists a number of different
ways to construct a path from a set of points, e.g. different interpolation techniques and
optimization (shortest route through a set of points, design for minimum time and energy, add
vessel speed and acceleration constraints). See for instance [1], [3] and [4.].
One of the most important aspects is that the reference signals to the control module is
smooth. If the constructed path is discontinuous, it must be smoothed in some way before the
reference signal enters the control loop. By initially creating a smooth path, we avoid
introducing additional complexity at a later stage.
We consider a parameterized path p parameterized by

North

East

Theta is denoted as the path parameter and p() gives the position according to theta in an
Earth-fixed reference frame

DP Conference

Houston

October 12-13, 2010

Page 3

Ivar Ihle

New Applications

Improved DP tracking usability

A set of waypoints and a path parameterized by length.

We will later show how such a path can be obtained from a set of waypoints, but we will first
discuss the motivation for using parameterized paths. For a thorough examination of
parameterized paths in reference models, see [5].
The parameterized path and its derivatives contain all information about the paths geometric
properties: position, tangent, tangential path heading, curvature, etc. Assume the path
constructed from the waypoints is given as

North

East

,
m
.

We use the following short-hand notations to define the path derivatives.

: d p

2
p

2
: d 2 p

p 3

3
: d 3 p

,
m/

2
y

,
m/2

3
y

,
m/3

and find the path heading, curvature and change of curvature from the path derivatives

DP Conference

Houston

October 12-13, 2010

Page 4

Ivar Ihle

New Applications

t tan1

Improved DP tracking usability

y
,
rad
x

xy x y
,
rad/
c
y 2
x2

x y x y 2c x x y y
3

x 2

y 2

,
rad/2

The parameterized path consists of smaller path segments and is constructed as follows

1
2
3

Path Construction Algorithm

Find suitable path segment endpoints based on a list of waypoint locations and
acceptance radius
Determine endpoint derivatives. Smoothness of the entire path follows by using similar
endpoint values for path segments before and after endpoint.
Find polynomial representation for path segment using a proper path parameterization
length.

The advantage with this path representation is that we are able to specify complicated paths
composed of a number of simple motions since initial position and curvature can easily be
matched with the previous segments position and curvature.
In many applications, the primary goal is for a reference point outside the vessel to follow a
path. This can e.g. be a plough or pipe section on the seabed. To account for this, we use a
vector l with distance from vessels centre of gravity to reference point, and then compute a
path for the vessel centre of origin/gravity (CO). When centre of origin is equal to the centre
of rotation (CR), we find the CR path as follows:

p CR
pR

lx
ly

R
x
where

cos x sin x
sin x

cos x

is the tangential path heading, and lx and ly is the longitudinal and lateral offset. R(x) is a
rotation matrix. The longitudinal offset lx is often referred to as the touchdown distance since
this would be where a e.g. a cable touches the sea bottom. The derivative of the rotation
matrix is

xR
x
S
x
, S
r

DP Conference

Houston

0 r
r 0

October 12-13, 2010

Page 5

Ivar Ihle

New Applications

Improved DP tracking usability

Centre of
origin/gravity

dx

dy
Centre of rotation

Touchdown
distance lx
Wpt path
reference point

Offset
distance ly

Waypoint path

Vectors between centre of gravity, centre of rotation and desired path location

The CR path derivates are

p CR
p RScl
p CR
p Rc2 RSl
2

In a similar manner, we find the path and path derivatives for the centre of origin when it
differs from the centre of rotation by using a vector d with distance between CR and CO in
the vessel-fixed reference frame:

p CO
p CR
R

dx
dy

where is the vessel heading. We find the CO path derivatives by substituting the curvature
c with vessels rate of turn.
The figure shows the three different paths and a vessel with fixed heading to better illustrate
the effect of vectors between the different paths
By replacing the derivatives for the waypoint path p() with derivatives for the CR and CO
path, we are able compute similar path information (tangent, curvature etc.) for both the CR
and CO path.

DP Conference

Houston

October 12-13, 2010

Page 6

Ivar Ihle

New Applications

Improved DP tracking usability

WPT path
CR path
CO path

80

60

40

20

-20

-40

-60

50

100

150

200

The waypoint path (green), CR path (red), and CO path (blue).

Path Speed Profile

Speed

We have now found the path that the vessel shall follow, but have yet to decide the progress
along the path. A tracking task nearly always involves some sort of dynamic behavior along
the path: usually, a time, speed or acceleration assignment must be met. The dynamic
assignment may be automatically generated, supplied with the waypoints, or may come from
an operator. Some examples of such assignments are:
keep constant speed along path,
maintain a steady speed for a certain time period,
be at a certain waypoint at a certain time,
complete track at minimum time without violating operation constraints, and
use a specified force for acceleration/braking

Speed profiles based on different dynamic objectives: Red curve: minimum time, blue: minimum
force.

DP Conference

Houston

October 12-13, 2010

Page 7

Ivar Ihle

New Applications

Improved DP tracking usability

We generate a path speed profile with a speed assignment as a function of covered path
distance. This is the speed to be kept on the waypoint path. The corresponding speed on the
CR and CO path is computed from the path derivatives. Consider the next figure: to maintain
a given speed on the original waypoint path the vessel must go faster during turns due to a
smaller path tangent value.
v [m/s]

Wheel-Over Point

CO path
Normal
offset

>v

Wpt path
v [m/s]
v

Speed Profile
Wpt path

Speed Profile
CO path

Speed profiles on WPT and CO path with an offset normal to the path direction

The information about the waypoint, CR and CO path and their derivatives let us:
Know properties of paths for centre of rotation, centre of gravity and waypoint table.
Verify path construction
Check that chosen offset is ok (depending on track construction strategy)
When we combine the path information with the speed profile we have more versatile tools
available that enable us to:
find speed and acceleration values on all path segments
analyze path speed profile (certain speed/acceleration values) against limiting factors
for entire waypoint, CR, and CO path.
detect potential conflicts between simultaneous operations, e.g., heading change
during a turn and keeping maximum speed on track
calculate new speed setpoints to eliminate conflicts
Factors that limit speed assignment include, but are not limited to, vessel speed in turns,
rotation during heading change, vessel speed/acceleration/rate limits, and dynamic limits for
the reference point along the waypoint path.
Solutions are found analytically and are thus available instantly, eliminating the need for
algorithms that involve searching and trial-and-error.
Example: Maximum path speed vs. heading change operation and vessel speed limits
To demonstrate the proposed method well consider a vessel in a heading change operation
270 degrees clockwise (for presentation purposes) during a 90-degree turn. The operation is
illustrated in the figure below.

DP Conference

Houston

October 12-13, 2010

Page 8

Ivar Ihle

New Applications

Improved DP tracking usability

The objectives in this operation are:

1. We want the waypoint (which, in this case coincides with CR path) to be followed
with a certain speed
2. The corresponding CO path speed shall stay below vessel speed limits
3. The heading rate-of-turn (ROT) must be large enough to finish heading operation.
These objectives may conflict, but knowing the CR and CO path derivatives, the vessel speed
limits in CO and heading ROT we are able to compromise on a the maximum CR path speed
that satisfies objective 2 and 3.
We consider the first two objectives: The speed limits are given as surge and sway values in
the vessel frame. We therefore convert the CR speed to the vessel frame. Given a CR path
speed assignment and path geometry, the inertial speed in CO is

V CO p CO V CR
We then have that

V vessel
R
V CO R p CO V CR
CO
Using this identity, we are able to check whether a speed assignment Vcr exceeds the vessel
speed limits, and by calculating backwards, use the speed limits to find the maximum speed
assignment.

160

140

120

North

100

80

60

40

20

0
-50

50

100

150

200

East

The path for the centre of rotation is shown in blue, and the path for the centre of gravity is red.
Different vessel snapshots show the operation progress.

Path parameter update law

We have now generated the path and found a feasible speed profile along it. We next look at
the update law for the reference model. The update laws task is to send reference signals to
the numerical control law such that the vessel moves according to desired path and speed.
The path parameter is the precise reference position on a path segment, and by constructing
an update law for we control the progress along the path.

DP Conference

Houston

October 12-13, 2010

Page 9

Ivar Ihle

New Applications

Improved DP tracking usability

A straight-forward solution to advance the reference model along the generated path would be

vp
where vp is the path speed at the current location. This will ensure that the reference model
follows the path speed profile and sends out matching position and velocity reference signals
to the controller.
However, to make sure that reference position and speed stay close to the vessel position and
speed we implement the path parameter update law with mechanisms that correct the path
parameter progress given the error e between vessel and reference model position:

fvp , e
Deviations between vessel and reference model occur because of unmodeled disturbances etc.
and e provides vessel feedback to the reference model. When there is zero deviation, the path
parameter follows the path according to vp(theta) and, otherwise, rapidly catches up with the
vessel position.

Operator Interaction
This section will give an overview of the user interface, discuss items that we think improves
usability and show examples of how the theory from the previous section is applied.
The user interface for waypoint tracking is implemented in the existing DP system and adopts
the existing graphics scheme. New graphic symbols must also match the existing to ensure a
familiar experience. However, since a tracking operation differs from a change position
operation, we have chosen to make parts of the user interface and input device operation
specific. That is, after a tracking operation has started, the graphical user interface will
emphasize information that is essential for tracking. The input device also adapts to the
situation with new commands replacing those that no longer apply in a tracking situation.
When an operator has the freedom to enter and manipulate e.g. waypoint data, probably
situations will arise where you get unexpected results, such as strange path shapes, conflicting
speed and heading setpoints etc. Restricting the operators possibilities reduces the chances of
unexpected situations, but affects user experience and may feel conservative. The tracking
user interfaces goal is to give the operator as much freedom of choice as possible and
preparing for potential operator blunders by:
1. Minimize chance of mistakes occurring
2. Minimize impact if mistakes occur
Errors can, and will, be made. Hints, instant and visual feedback on errors can show the user
what went wrong and why the situation occurred. The goal is to both minimize impact and
explain and educate the user such that errors can be avoided in the future.
A waypoint tracking operation starts by loading the waypoints into the waypoint table. From
here you can make changes to values for each specific waypoint. You can also save and load
a track from and to a file, rename your track or load a new track from a chart system.
Whenever a change has been made to a waypoint, a yellow marker will appear in the
Applied column. If the change made resulted in an invalid track, the marker will be red.
Path analysis methods and choice of path construction strategy determine the track validity.
Changes are approved by pressing the Apply button.

DP Conference

Houston

October 12-13, 2010

Page 10

Ivar Ihle

New Applications

Improved DP tracking usability

The waypoint path is constructed from the waypoints and shown on the 3D scene as a yellow
path. It represents the predicted path of the vessel centre of rotation. After the operator has
activated the tracking operation, the waypoint track is shown as a grey highway. The green
box, located at the centre of the vessel, visualizes the cross track error limits.

Waypoint table

Note that the heads-up-displays (HUDs) change after the tracking operation has started. The
HUDs supply the user with what we have identified as key knowledge for the current
operation. During a tracking operation the user sees information about track progress, speed
setpoints in turns and on straight line segments, track deviation and chosen offset values.

For a constant speed profile, the reference speed in turns may vary from straight line
segments due to path curvature, heading change operations and vessel speed limitations. The
maximum allowable speed through a turn is calculated in a similar manner to the example in
the previous section.

DP Conference

Houston

October 12-13, 2010

Page 11

Ivar Ihle

New Applications

Improved DP tracking usability

We aim to incorporate a what-you-see-is-what-you-get principle in the user interface.

Changes to the waypoint geometry or offset vector are shown instantly as a yellow edit curve
on the 3D scene and provides visual feedback. During tracking operations, the transition
from the current highway to the new edit curve is always shown. The shape depends on the
current along-path speed and the selected speed towards the offset track (perpendicular to the
track direction). After applying the changes the yellow edit path becomes the new highway to
be followed.

Path with new offset

Path analysis from the previous section makes it easy to quickly analyze changes to the path,
speed profile and other operational changes. If the operator makes changes that cause an
invalid path (e.g., two waypoints are too close, new offset value is too large), the edit path
will be shown in red and it will be impossible to apply the changes. In addition, the
waypoint(s) with invalid settings will be marked in the waypoint table.

Path with moved waypoint

DP Conference

Houston

October 12-13, 2010

Page 12

Ivar Ihle

New Applications

Improved DP tracking usability

Transition from current to new path

Most changes and commands are performed in the Tracking Operation pane. However, three
settings, manipulating the perpendicular offset distance and adjusting the speed and heading
setpoint, are also available from the input device during low-speed tracking operations.
Interviews with DP operators identified these as the most used commands and making them
more accessible lightens the DP operators work load.

Some situations must of course be handled manually. The way-point tracking will go into a
fault state if one of the following conditions occurs:
1. All position reference systems lost (causes dead reckoning).
2. All heading sensors lost, (will also cause dead reckoning).
3. No thruster forces in surge, sway or yaw.

DP Conference

Houston

October 12-13, 2010

Page 13

Ivar Ihle

New Applications

Improved DP tracking usability

When going into the fault state the controller will try to stop the vessel: speed set points are
set to zero and when model estimated speed has reached zero, the controller will try to keep
the estimated position.
When thruster forces are available in 3 DOF and position reference systems and heading
sensors are OK the system leaves the fault state. If the vessel has drifted of more than 3 meter
from last known position on track (or more than 3 degrees), a dialog box will pop up with
two selections, continue or exit tracking. The yellow curve will appear and show the route
back to track. The operator must then decide what to do next.

Conclusion
We have seen how to construct a parameterized path from a set of waypoints and how to
apply it in the reference model of a low-speed waypoint tracking system. The path provided
information about position, path tangent, curvature, and in combination with a speed profile,
we determine speed and acceleration values for the entire path. All these values can also be
found for the path described by the centre of rotation and centre of gravity. Various usages of
this information to improve usability were discussed. Parts of the user interface were
presented: operation specific information available when tracking operation starts, fast path
analysis, visual feedback on path geometry and setpoints. When combined, these elements
reduce operator workload during waypoint tracking operations.

References
[1] Fossen, T. I., Marine Control Systems, Marine Cybernetics, 2002.
[2] IMCA, Introduction to Dynamic Positioning, http://www.imcaint.com/divisions/marine/reference/intro.html. Online version: August 18th 2010.
[3] Sciavicco, L. and Sicilliano, B., Modelling and Control of Robot Manipulators,
Springer, 2000
[4] Skhel, Lumelsky: Classification of the Dubins set, Journal on Robotics and
Autonomous Systems 34 (2001) 179202.
[5] Skjetne, R., The Maneuvering Problem, PhD thesis, Norwegian University of Science
and Technology, 2005.

DP Conference

Houston

October 12-13, 2010

Page 14