16.03.2021Blog

Towards sustainable, smooth and safe marine transport with artificial intelligence

Over 80% of the Finnish foreign trade is carried by sea. The supply chains have become increasingly complex and valuable. At the Nordic waters, winter sets extra challenges for maritime transport. In this story, we take a look at how we started to advance towards using artificial intelligence in AWS for just-in-time marine transport at the Nordic waters covered with ice.

Everything you see at the moment was most likely carried to you by maritime transport. The screen that shows this text, the clothes you have on, practically everything has been transported by a merchant vessel at some point in time in a complicated, international logistics chain. According to the statistics by the Finnish Customs, some 80-90% of the Finnish foreign trade is carried by sea. Marine transport is expected to even grow according to this report by the United Nations in 2019.

In Finland, the winter is coming. Every year. The seas are covered with ice. While the average number of days of sea ice coverage at the baltic sea is decreasing to 100-150 due to climate change, the Northern Sea Route (NSR) is opening up. There were 18 icebreakers assisting vessels passing NSR last summer. In Finland, we have several icebreakers operating at the Baltic Sea at the moment. A typical setting is that an icebreaker leads the vessels in an efficient convoy.

The distance between the vessels should be short since the ice field closes very quickly. If the distance is too long, an assisted vessel might get stuck into the ice. On the other hand, a very short distance is dangerous due to the risk of collisions that happen every winter in the Nordic waters.

The following map animation captured from baltice.org shows how the icebreaker fleet works when a vessel suddenly gets stuck into the ice.

A merchant vessel (blue circle) approaches the ice field from the south. Suddenly, it gets stuck into the ice. The icebreakers (orange circles) assist the vessel and it is able to continue its voyage towards the destination port. Without icebreaker assistance, the northern ports would be closed during the wintertime.

AI to predict sudden stops of the vessels

In March 2018, we conducted an AI experiment with the Finnish Transport Infrastructure Agency (FTIA) based on open data in the Digitraffic service in co-operation with the University of Jyväskylä. As you may know, even 80% of all work done in AI experimentation is often about data cleaning, preprocessing and other data engineering tasks. That’s exactly what we did to train a regression model. We mined vessel location data, combined it to satellite images, and reconstructed one of the collisions at the sea. The question was: how could artificial intelligence (AI) help in predicting sudden stops of the vessels? The icebreaker fleet could benefit from such new information. It could:

  • optimise icebreaker movements
  • shorten extra waiting times of the vessels at the sea

To put it short, AI could help to make marine transport a bit more sustainable, smooth and safe. The intermodal logistics could be optimised better. For instance, road hauliers would benefit from better situational awareness at the sea. They could arrive at the port just-in-time. Moreover, a train transport can have a schedule where the train leaves from the port every other week.

The tool we used in the experiment was Amazon SageMaker. We imported the vessel location data into the tool. The condition data, for example, wind forecast and ice forecast, could then be combined with the vessel meta and location data on top of satellite images.

To hear more about the experiment, please join the Amazon Web Services (AWS) In Spotlight webinar to hear how artificial intelligence and machine learning helps to accelerate your cloud journey, allow you to create innovative services, and deliver better customer experiences. Neil Mackin, technical business manager at AWS, and Timo Lehtonen, senior software developer and an RDI data scientist at Solita Oy will share the latest trends, insights, and customer use cases.

The work has been done in two Solita Research projects, namely Gateway to Mercury Business and 4APIs. Solita Research is a continuum of Solita Science that was executed in 2015-2019. We have several ongoing research projects funded by Business Finland. In the future, we’ll bring the human into the loop as part of the AIGA project. Feel free to contact the author to hear more about Solita Research opportunities for you!