ries
barette
d'imagerie
1ation des
inventaire
chacune de
S. et. âà.-la
r qu'elles
égrées aux
(SIG). et
raite d'un
érencement
barette de
rpore des
provenant
rigation a
a position
-imageurs.
ffectue la
L.gorithmes
eres. On
le données
acité de
ité de la
ée par une
sciences
GUIDING SHIPS ON WATERWAYS
M. Sandler, U. Kabatek, R. Neul, E. D. Gilles
Institut für Systemdynamik und Regelungstechnik, Universitát Stuttgart
Commission ll, Working Group 1
KEY WORDS: ship guidance, radar image processing, sensor integration
ABSTRACT
A self-contained, integrated navigation system for the automatic guidance of ships on waterways is described.
Besides the usual navigation sensors, imaging sensors like a radar or a laser scanner are used. The sensor
information is combined with a-priori knowledge in the form of an electronic chart of the waterway and a dynamic
ship model, in orderto determine the position and heading of the navigating ship in global coordinates. Matching
of image data and electronic chart is performed by a least-squares matching technique that is applied similarly to
measurements from the different imaging sensors. The results of the matching processes and measurements
from other sensors are integrated by a Kalman filter. A multiple-target tracking method is implemented for
the determination of the trajectories of other vessels and thus for the evaluation of the actual traffic situation.
Additional processing steps of the integrated navigation system are the planning of a trajectory and the
guidance of the ship along this trajectory. The methods were successfully tested in practice. The structure of
this integrated navigation system, developed for the case of inland shipping, may be transferred to other traffic
systems.
1 INTRODUCTION
At the "Institute for System Dynamics and Control",
University of Stuttgart, an integrated navigation sys-
tem for inland and costal waterways is being devel-
oped. The goal of the project is a system capable
of automatically guiding a ship on a waterway in nor-
mal traffic situations. Such a system will relieve the
navigator from tiring routine work and support him
in complex situations, particularly at night or in foggy
weather. The heavy traffic on e. g. the river Rhine with
shipment of dangerous goods such as chemicals and
refinery products implies many risks for the environ-
ment. The system helps to protect the environment
by increasing the level of safety on the waterway. It
is not targeted as a replacement for the navigator.
This project is funded by the "Deutsche Forsch-
ungsgemeinschaft" within the SFB 228 'High preci-
sion navigation'. It is also promoted by the german
ministry for traffic.
Within this paper we will focus on the real-time pro-
cessing of the measurements of the different sensors,
especially the processing of the image data. Image
data are generated by a radar and by a laser scan-
ner both sensing the surroundings of the ship. 1n
section 2 the configuration and components of the
navigation system will be explained first. The elec-
tronic chart is an important component in the image
processing algorithms. Her structure and implemen-
tation is discussed in section 3. The image data are
used to determine the position and heading of the own
vessel in global coordinates by matching the images
to the electronic chart of the waterway. The match-
ing algorithms are discussed within section 4. As
explained in section 5 the image data are also used
to determine the actual traffic situation. Based on the
matching results and measurements of other sensors
available to the system an integrated estimate of the
ship's state is computed by a Kalman filter described
in section 6. In section 7 the generalized structure
of the navigation system is deduced. Finally section
8 gives a brief outlook on the trajectory planning and
control algorithms implemented within the navigation
system.
2 COMPONENTS AND TASKS OF THE
INTEGRATED NAVIGATION SYSTEM
The configuration of the integrated navigation sys-
tem is shown schematically in figure 1. Onthe left side
the sensors of the system are represented. Imaging
and non-imaging sensors can be distinguished. The
radar yields a map-like representation of the local en-
vironment. Itis part ofthe standard equipment of most
commercial inland and coastal ships. A new radar im-
age is obtained about every 2.3 seconds and trans-
ferred into the ship-borne computer. Processing of
the radar image is very computationally intensive and
has to be accomplished in real-time. Therefore, all
algorithms for this task have to be designed for max-
63
