After three years of hard work
the AI-ARC
("Artificial Intelligence based Virtual Control Room for the Arctic")
EU-funded research and innovation project concluded in February 2024.
Started in 2021, the project’s main objective was to create
an innovative and user-friendly AI based platform, the Virtual Control Room,
aiming to improve maritime situational awareness, decision-making, communication, and thus safety of all maritime actors,
particularly in the Arctic Sea.
The project was implemented by a
consortium of 22 partners from 12 countries, coordinated by Laurea university
(Finland) and involving as end users the Icelandic Coastguard, the Maritime and
Coastguard Agency (UK), the Swedish Coastguard, the Joint Rescue and
Coordination Centre Northern Norway and the Department of Defence – Irish Naval
Service.
The project developed a suite of innovative
services:
- Prediction
of Icepack & Iceberg Movement
Prediction of icepack and ice
cover along shipping lanes using weather information, radar, optical images and
maritime traffic data. Prediction of the need for icebreaker aid in an area,
via ice status of shipping lanes and icebreaker movement.
Iceberg track prediction uses a
deep neural network to analyse historical track records from satellite images,
together with weather and sea conditions.
- Vessel
Traffic Prediction Service
Vessel movement prediction is
carried out over various timeframes using a Machine Learning approach on AIS
data, which is specifically designed for rapid processing. Deviations from
predictions by actual ship movements are detected and heatmaps of historical
vessel movement events are generated, over the sea area of interest.
- Search
and Rescue – Wide Area Search
A service for SAR authorities to
predict the drift track of a missing vessel, combined with dynamic search
pattern updates to the SAR vessel bridge. It is based on last known position,
weather and ocean current data, combined with satellite-based image recognition
data.
A suite of services based on a
variety of anomaly-detection methods – AI, ML & deep learning – to identify
misalignments, deviations and outliers in the behaviour of vessels, using their
positions and trajectories.
Automated detection of oil spills
or other environmental emergencies based on processing Earth Observation
(Satellite) data. Results are visualized in the AI-ARC platform, enhancing the
probability of detecting unexpected phenomena.
- Satellite-Based
Sea Ice Coverage Maps
Sea ice concentration maps are
derived from CMEMS (Copernicus Maritime Environment Monitoring Service) daily
data. Maps are provided via the AI-ARC system to aid safe navigation and route
planning.
- Risk
Index Computation Service
Provides end users with a visual
5-level risk indication for navigation based on a multi-criteria model,
dynamically updated, based on vessel data, meteorological and ocean conditions.
- Reliability
Assessment of ML-Services
A service to assess AI models
based on their reliability in terms of the technical aspects: transparency,
performance, and robustness.