IEEE Robotics & Automation Magazine - September 2023 - 11

DATA COLLECTION AND TRANSMISSION
With the growth of emerging demands from marine applications
such as seabed survey, pipeline maintenance, and wind
turbine installations, the main concern for the maritime
industry and shipowners is how to achieve efficient marine
operations while ensuring safety. Today's maritime engineering
systems are often equipped with various sensors and
operate in highly dynamic environments. Incorporating sensor
data not only about the internal status of machinery, propulsion,
and engine systems but also from cameras, lidar,
radar, sonar, and GPS sensors improves the situational awareness
of the ship. Figure 3 illustrates how to use historical/live
data and cosimulation technology to achieve a digital twin
system. All of these data sources, including status from the
ship, crane, and engine, can be transferred via a 4G connection
to an onshore control center to conduct operational
planning. A Next Unit of Computing PC is deployed on Gunnerus,
from where the Message Queuing
Telemetry Transport network
protocol is used to publish data messages
from different data sources on
demand. Then, a local server is used to
subscribe to each data source to store it
in a database. Finally, the data are
applied and analyzed on the digital
twin and transferred back to Gunnerus
to enhance operational planning and
decision making.
User
Interface
MODELS AND SUBDOMAIN
MODELS FOR COSIMULATION
The models and submodels that represent
physical systems and behavior
are the building blocks of a functional
digital twin. These digital representations of real systems and
behavior can be physics- or data-based algorithms or only
graphics for visualization purposes.
The cosimulation approach for complex systems, like a
ship, has its advantages in terms of modularity, modeling flexibility,
and simulation efficiency. However, it always depends
on the application to determine complexity of the models to
ensure both adequate simulation accuracy and efficiency. For
example, as shown in Figure 1, in the ship-design phase, relative
low-fidelity models will be enough for fast prototyping to
make a quick whole-ship concept test. After that, we need a
high-fidelity model for detailed design, control, and prediction.
For the training phase, real-time models will be required.
The balance between model complexity and efficiency has to
be considered.
As mentioned earlier, the cosimulation framework
allows for individual step sizes for each subsimulator and
Visualization
Monitoring and
Control System
Master Algorithm
Standardized Cosimulation Interface
(FMI)
OEM A OEM B OEM C
Remote
Interface
Sensor Data
OEM D
Digital Twin
Components
FIGURE 2. The OSP. OEM: original equipment manufacturer.
Onboard Control
Center
* Efficiency
* Safety
* Environment
Operational
Decision Support
* Remote Operations
* Autonomy
* Dispersed Teams
* Expert Assistance
Onshore Control
Center
" Digital Twin "
Data Storage
Data Analytics
FIGURE 3. Data collection, transmission, and storage.
SEPTEMBER 2023 IEEE ROBOTICS & AUTOMATION MAGAZINE
11
Data Transfer
Data Transfer
Operational Planning
Data Transfer
IEEE Robotics & Automation Magazine - September 2023

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