IEEE Robotics & Automation Magazine - September 2023 - 10

MARITIME DIGITAL TWIN SHIP
The RV Gunnerus [10] from the Norwegian University of
Science and Technology (NTNU) is used as a testing platform
for our research, as shown in Figure 1. It is equipped
with the latest technology for a variety of research activities
within biology, technology, geology, archaeology, oceanography,
and fisheries research. The main dimensions of the vessel
are given in Table 1.
The digital ship can either be deployed on its physical counterpart
or placed on shore in a remote control center, given that
the sensor data are able to be transmitted via a gateway. A historical
data set can be generated by taking all-year measurements
from the onboard sensors placed on the physical ship,
e.g., the ship's hull, engine, and thruster. Digital models of systems
and subsystems in simulation can thus be refined using
combined model- and data-driven approaches [11] to ensure
fidelity of the digital twin of a ship. Based on the digital ship
simulation, a variant of control levels, from remote control to
fully autonomous, can be implemented in
the digital twin system. As a result, different
onboard support tools, such as visualization,
task configuration and high-level
applications like path planning can be
included in the system. The result of the
onboard support tools will either be fed
into simulation for validating operations
or be treated as metrics for decision making
on the physical ship.
Although the implementation of the
ship with high fidelity can thus be realized and used as
digital tools for either decision making in remote control
center, or control basis for an autonomous ship.
Four key issues in a Gunnerus digital twin will be introduced
separately, including
■ cosimulation mechanism as digital twins platform
■ data collection and transmission
■ models and subdomain models for cosimulation
■ enabling tools for onboard support on the ship or on shore.
"
digital twin has some current limitations,
such as insufficient possibilities
for synchronization between the physical
and digital worlds, the lack of high-fidelity models for
simulation [5], [12] as well as challenges for gathering and
processing large data sets (Figure 1 shows a sound conceptual
framework) from which a digital ship can be refined
via historical/onboard sensor data using data-driven methods
and be combined with other models used in marine
operations for onboard support or advanced control. As a
sequence of evolution from classic simulation, to virtual
prototyping, then to digital twin implementation, a twin
TABLE 1. The main dimensions of Gunnerus ship.
PARAMETER
Length overall
VALUE
Length between perpendiculars
Waterline length
Breadth middle
Breadth extreme
Depth middle main deck (Dm)
Draught middle
Deadweight
36.25 m
33.9 m
29.9 m
9.9 m
9.9 m
4.2 m
2.7 m
165 t
THE OSP CONSISTS OF A
COSIMULATION LIBRARY
WRITTEN IN C++, WITH
ADDITIONAL INTERFACES
FOR C AND JAVA.
„
COSIMULATION MECHANISM AS A
DIGITAL TWINS PLATFORM
Cosimulation is a promising technology that enables different
subsystems to be modeled and simulated in a distributed
manner [12]. In general, it is difficult to apply models implemented
in different tools and domains into one simulation.
However, with the emergence of two noteworthy standards,
namely, high-level architectures and the FMI, different subsystems
can be modeled separately and
composed into a global simulation where
each model is executed independently,
sharing information only at discrete
time points. In this work, only the FMI
is considered. However, to effectively
make use of and connect different subsystems,
some higher form of orchestration
layer is required. Several such
orchestrators exist, both open source
and commercial, each with its pros and
cons. An overview of various open source
solutions is provided in [12].
We rely on the open-simulat ion
platform (OSP) [13], which is specifically designed for the
maritime industry for performing cosimulation and sharing
simulation models, e.g., Functional Mock-Up Units (FMUs).
The OSP architecture is shown in Figure 2. Thus, a ship can
be implemented as an aggregation of several independent submodels,
including the hull, thrusters, and power system, and
so on.
The OSP consists of a cosimulation library written in
C++, with additional interfaces for C and Java. A simple web
GUI and command-line interface is also available. The OSP
has developed the OSP-interface specification (OSP-IS),
an addition to the FMI that provides a method for adding
semantic meaning to model-interface variables. Additionally,
proxy-fmu [9] will be used to enable cosimulation of
otherwise incompatible simulation models, and the System
Structure and Parameterization (SSP) standard will be used
to facilitate configuration of the overall cosimulation structure,
including its parameterization and connections. To
simplify the process of creating SSP-compliant systems, the
project partners have developed SSPgen, which allows such
systems to be defined using a domain-specific language that
also enables the OSP-IS to be applied in an SSP context.
Currently, OSP is valid for the public. Detailed information
can be found in [13].
10 IEEE ROBOTICS & AUTOMATION MAGAZINE SEPTEMBER 2023
IEEE Robotics & Automation Magazine - September 2023

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2023
