The maritime industry is amid a digital transformation, but there is still a long way to go. Moving forward, we must create smart systems that improve upon the way humans understand and act when faced with various scenarios. For this, data is needed, and the industry has already collected immense amounts of data over many years. The right toolbox with enough computational power is necessary to bring it all together.

Vessel AI (Artificial Intelligence) is an EU-funded research project and organization, consisting of maritime stakeholders coming together as industrial, technical, and academic partners to create a common framework for gathering and interpreting data that is to be shared openly with the global maritime community. Using artificial intelligence, machine learning and HPC (High-Performance Computing), the group aims to facilitate maritime digitalization, making the sector smarter and more sustainable.

Route prediction

As a Vessel AI partner, Kongsberg Norcontrol will use technologies developed in the project for accurate ship route prediction. AI-based models are trained using historical and real-time AIS (Automatic Identification System) data. Combining that with data from SafeSeaNet, Norway's national ship reporting system for vessels to report their arrival and departure information to the Norwegian ports and government agencies, the system can predict where the ship will be at any given time.

Thorbjørn Tønnesen Lied, PhD. Principal Software Development Engineer at Kongsberg Norcontrol

Having a long-term perspective on where ships are and will be is incredibly useful, especially for shore-based operators, who need to know what the traffic picture will look like in the future, to plan ahead and avoid unwanted situations. Should a ship act in an unexpected way, for example, it might be a sign of problems that could lead to a collision or grounding. It could also be an indication of illegal activity. Either way, the system will flag the incident and alert the shore-based operator about deviations from the expected route. The operator can then assess the situation and decide on the best course of action.

Another use for route prediction is resource planning for congested traffic centres, such as ports, canals, and other narrow inland waterways. Too often ships sail at full speed to reach their destination as soon as possible yet are often forced to wait to get through areas with limited capacity.  This practice, which wastes both time and fuel, can be avoided if ships know in advance the optimal route and speed at which to sail. A route prediction service can consider factors such as traffic (historical and real-time), bathometry, and sea and weather states to optimize the vessel’s voyage for a just-in-time arrival. Being able to determine the optimal speed and route is also highly relevant for autonomous vessels who will use pre-defined routes.

Intelligent decision support

Kongsberg Norcontrol is a leading provider of maritime domain awareness solutions. Our system presents the operator with traffic image or common operating picture to improve the operator’s situational awareness. We are currently in the process of evolving that system beyond the real-time vessel traffic image into a proactive, forward-looking system that can both show the current traffic image as well as predict what will happen in the future. By developing services such as Route Prediction, we can deliver accurate decision support for the operator to make more informed decisions than they would on their own. This in turn will allow for safer, more efficient ship traffic where there are fewer accidents and less emissions.

The Route Prediction model is currently in a development phase but will be put to test in Vessel AI Pilot 3, which is called “Autonomous Ships in Short Sea Transport”. In this pilot, we are delivering the software solution to be deployed at the newly established SCC (Shore Control Centre) in Horten where our partner Massterly operates two autonomous ASKO cargo ships from. Using the algorithms developed in Vessel AI, we can predict beyond the next manoeuvre of the other ships travelling in the Oslo fjord, finding the best route and time slot for the autonomous ships to pass.