Home Bots & Business Case Study: Finland takes a big step to digital shipping with a new national machine learning system

Case Study: Finland takes a big step to digital shipping with a new national machine learning system

by Gary Flood

Finnish policymakers think use of a new ‘time data’ system based on tech from a local specialist AI maritime logistics company, Awake.AI, can do everything from improve the competitiveness of the nation’s entire maritime logistics to aiding the environment and meeting a national goal of “efficient and sustainable logistics by means of digitalisation”

A new service that uses machine learning to calculate the estimated arrival times of merchant ships to all ports in Finland is about to be launched.

Put like that, maybe not so exciting—but this new system is set to radically improve operations in one of that nation’s most important industries and is set to do everything from boost the efficiency of port operators’ routine activities, speed up automation, help to anticipate exceptional circumstances and reduce environmental emissions.

The new software is being rolled out by the Vessel Traffic Services arm of Fintraffic, a state-owned company that controls and manages Finnish land, air and sea traffic by providing data that helps companies create new traffic and smart mobility solutions for people and goods. Earlier this month, a local Finnish AI company focused on maritime logistics, Awake.AI Oy, has confirmed it has won the EU-wide tender for the proposed ‘time data’ service for port operators and authorities, a.k.a the new ‘Port Call Time Stamp and Estimation Service’.

‘Better information sharing was required to boost operational efficiency’

“There have long been attempts to tackle the challenges posed by data management in maritime logistics,” states Katariina Kalatie, Chief Advisor, Ship Technology and Marine Environment, at the Finnish Transport and Communications Agency, Traficom. Since 2019, she and her team have been working on the problem of improving shipping efficiency in the Scandinavian country. What emerged from their research: better information sharing was required to boost operational efficiency in the sector. However, some of this information contains trade secrets and many operators are in competition with each other, so this was not a straightforward problem.

Sharing of time data was eventually identified as something that would benefit all port operators equally. Specifically, each arrival and departure of a ship from a port affects the schedules of many travellers and businesspeople, and although most vessels can usually provide an estimated time of arrival via radio, the data obtained from the vessels has been poor and fragmented, while getting an accurate time is always influenced by a ship’s speed and route, weather and ice.

Now, using machine learning, all forecasts will be analysed using what had been a neglected resource: past AIS (Automatic Identification System) messages, the standard in commercial shipping. (Jussi Poikonen, Awake.Ai’s technical project manager: “The AIS messages sent by vessels hold a lot of potential, but for some reason their benefits have so far been ignored.”)

As a result of turning to these and other data streams, the new time data service will ‘remember’ past voyage timing and how various factors affected arrival time, plus learn normal deviations for a particular locality and adjust estimates, accordingly, reducing data fragmentation and generating more accurate schedules. APIs will also be generated from which port operators can obtain time data for use in their own systems. The new AI-based service will be introduced this autumn and, after a test phase, be customised for Finnish needs then potentially be offered to other states, say the partners.

Ports now ‘better able to plan their operations’

Fintraffic´s Vessel Traffic Services (VTS), which operates under the country’s Ministry of Transport and Communications, and which provides vessel traffic services to merchant shipping and other marine traffic as well as maintaining safety radio services, will oversee implementation. Its Programme Manager, Olli Soininen, explains why the system is so revolutionary:

“Estimated time of arrival (ETA) of merchant ships into a port has been identified as key in enabling more efficient planning of port operations, reducing waiting times and cutting CO2 emissions through traffic optimisation: ETA information is also something VTS needs as background information for traffic control as well as for the departure permit of vessels. With the introduction of the service, ports have learned to utilise information and, based on high-quality time information, are better able to plan their operations in relation to the arrivals of ships. And at the same time as operations become more efficient and streamlined, financial savings will be achieved by reducing waiting times and reducing emissions with better forecasting.”

Soininen also told us that arrival time information estimate is not the only data created; the service also generates actual time information events, that’s to say also lists when ships reached their destination versus the original estimate. Departure time, interestingly, can’t be estimated in the same way as arrival, as it requires situation information about the port operator’s loading situation which can often be sensitive, so it is not possible to share it the same way as the ETA.

In any case, the move marks a significant global first for both AI in shipping and the slow push to digital supply chains. As Soininen predicts, “The importance of time data for shipping has been taken into account in both national transport system plans and the government’s decision in principle on the digitalisation of logistics.

Misschien vind je deze berichten ook interessant