Wednesday, March 12, 2025     [ login ]

research

OCEAN project: a collaborative data sharing platform supporting enhanced situation awareness to reduce the risk of collision during navigation

Published: 14/10/2024

Ferry in Corsica. CC by Myrabella

Photo: Myrabella / Wikimedia Commons / CC BY-SA 3.0

Around 3,000 maritime incidents occur every year in the European maritime fleet. 28% of these accidents are categorised as severe or very severe accidents.

CIMNE is working with twelve other European partners to address maritime safety head-on through the OCEAN project (Operator-Centred Enhancement of Awareness in Navigation), an international initiative focused on enhancing operator awareness in navigation.

Experts from CIMNE's transport innovation unit, CENIT, are involved in this effort to reduce the frequency of serious accidents such as collisions and groundings, to protect the lives of marine mammals and to reduce the risk posed by floating obstacles. Through a transversal approach, OCEAN will promote safer practices by improving the usability of instruments and creating a Decision Support System for mariners.

Using human-centred design, which prioritises people’s needs and behaviours in product and service design, the project will utilise cutting-edge technology, such as AI algorithms, to enable sea crews to make better decisions, improving safety and protecting the marine environment.

The project aims to enhance operational efficiency and protect marine life by adopting advanced technology such as hydrophone grids, satellite imagery analysis and prediction models for both marine mammals and container drift while addressing many critical aspects including:

  • Navigational accidents root cause assessment

  • Enhanced Professional Standards for Situation Awareness

  • Human-centred Bridge Equipment Design Methods

  • Detection and tracking of Marine Mammals in high-density areas

  • Tracking and Recovery of Floating Containers

  • Improved Situation Awareness and Manoeuvring Planning

  • Automated Anti-collision Solutions & Shipboard Infrastructure


European Navigational Hazard Infrastructure

In addition to streamlining operations and protecting marine life, the project seeks to build the so-called European Navigational Hazard Infrastructure (ENHI), a system with the capability of receiving data concerning observations, detections and predictions from multiple data streams and able to optionally share part of this data with other components of the data ecosystem. This way, interfaces and formats to exchange data have been defined to support multiple sources. Mechanisms have also been developed to manage data in a proper way in terms of consistency, persistence and quality. Overall, from a design perspective, the ENHI can be seen as the sum of all the components depicted in the following diagram:

  • I/O Interfaces: Input and output interfaces defining how data is received and sent between the components of the OCEAN ecosystem.

  • Data Storage: Implements the underlying information system used to store both active data and historical information (expired data).

  • Management: Includes all the rationale implementing the use cases as well as supporting features concerning the control of accuracy, consistency and uniqueness.

  • Publishing: Feature used to build Navigational Warnings before they are published/delivered to the end users.

 

The image below illustrates the envisaged workflow in the OCEAN data ecosystem. As a general overview, reporting data is sent to the ENHI which, depending on the configuration of the system, can optionally request additional data to two on-demand components in charge of detecting containers or marine mammals through the analysis of satellite imagery or predicting the drift of a given container. The second half of the project (from April 2024 on) will address the support to the generation of Navigational Warnings (NW) to be delivered to the ships through existing communication channels as well as the incoming data product based on the International Hydrographic Organization S-100 framework specification.


For a better understanding of the whole system, the following list briefly outlines the most relevant features of the involved components:

Passive Acoustic Monitoring (PAM)

 

  • System developed in charge of detecting and locating whale audio

  • Composed of 4 hydrophones offering flat frequency receiving responses over a wide frequency range and relatively high transmitting sensitivity, making the solution very useful within many areas of underwater acoustic research, tests and measurements

  • Data collected is processed (beamforming using distance between hydrophones) to create 3D representation/spheroid of the surrounding space

  • Detected sounds are filtered and classified according to specific species by means of machine learning algorithms and, if necessary, sent to the ENHI

 

 

Reporting app

 

  • Progressive Web Application (PWA) targeting small craft and leisure market (while 4D-SAD targets SOLAS ships), whichallows users to report sightings of both marine mammals and containers as well as displaying Navigational Warning information

  • Reporting is possible at any time and data is sent to the ENHI whenever an internet connection is available

  • PWA technology allows for device-independent installation

 

 

Environmental Niche Models (ENM)

 

  • Statistical prediction of the probability of occurrence (or habitat quality) for some marine mammal species based on variables like sea surface temperature (SST), productivity, topography or oceanographic structures such as thermal fonts, among other covariates

  • The models run periodically to adapt results according to changes in the dynamic covariates of a specific geographic area

  • Output is given by means of a raster file with coloured pixels representing the probability/habitat quality. Then, a threshold is used to create polygons representing areas of high probability of animal presence

 

 

Satellite Imagery Analysis

 

  • Optionally triggered by the ENHI upon the reception of a report

  • A request is sent for a given area/time interval and both marine mammals and containers can be eventually detected after processing available imagery from the provider and sent back as additional data that complements the former report

  • After the application of pre-processing steps (geometric, radiometric and atmospheric corrections, pan sharpening), species-specific machine learning models (both for classification and object detection) are used for the actual detection of marine mammals

 

 

Container Drift Model

 

  • Regarding container losses and floating objects, this component is used to predict potential drift along a specific time interval considering the timestamp and location of the event that triggers the request and the corresponding weather and sea conditions like wind or wave height, for example

  • The output, based on the results of simulated data using OpenDrift, is then converted to geographical areas and sent back to the ENHI in GeoJSON format

  • In the same way as the satellite imagery analysis, it is integrated in the data ecosystem as an optional on-demand component triggered by the ENHI

The ENHI gathers near real time data concerning hazards from many different sources and provides mechanisms to generate NW that can be used to feed mariners with valuable information in order to replan the routes in advance or support manoeuvres during the voyages with the objective of avoiding collisions.

The OCEAN Project is an innovation and development initiative co-funded by the European Union. UK participants are supported by UK Research and Innovation Grant Number 10038659 (Lloyd’s Register) and Grant Number 10052942 (The Nautical Institute)



Adaptation of an original article written by CIMNE’s Francisco Miguel Rodero Blanquez


research  research for society  EU project