Skip to content
EN
Contact

AI in ports and the maritime sector

We start with a single decision about a quay, berth or waterway, using the AIS and port data you already have, as with Port of Amsterdam.

Discuss your port challenge Read the Port of Amsterdam case study

We gave Port of Amsterdam insight into historical berth usage

Space in the port area is limited. Together with Port of Amsterdam, we developed a tool that measures historical berth usage, so policymakers can use data to decide whether to provide more or fewer berths for certain types of vessel.

Read the case study

Do we know how our berths and waterways are actually used?

The harbour service can see which vessels are approaching in the AIS data, but whether capacity for a particular type of berth is too tight remains an estimate. By linking AIS positions to berths and notifications, you can see how often and for how long vessels berth at each quay.

Book a meeting
Binnenvaartschip ligt aan een ligplaats met meerpalen in een rivierhaven onder een blauwe lucht.

We make maritime data useful for planners and policymakers

Discuss your port question

Berth use by the numbers

For Port of Amsterdam, we built a tool using Databricks, real-time data and Power BI to measure historical berth use.

AIS data in a model

For Rijkswaterstaat, we combined AIS data with a congestion model and an LSTM to simulate how a bridge opening affects vessel queues.

Live data from the water

With DutchSail, we used Kafka and Grafana to develop a real-time dashboard that lets coaches track steering behaviour and sail trim in the simulator.

Monitored after delivery

A model becomes outdated when vessel traffic patterns change. For Rijkswaterstaat, we built a package that monitors models in production for drift and outliers.

Every project starts with the port call, from arrival to departure

  1. Decisions and data sources

    Which decision about berths, locks or scheduling do you want to support, and what AIS, reporting and scheduling data is relevant?

  2. Reconstructing port calls

    We link AIS positions to berths and reports, so each port call has an arrival, time alongside and departure.

  3. Insights first, predictions second

    We start with a dashboard based on historical data. We only build a model if forecasting adds value.

Is our maritime data ready for a model?

Talk it through with us

A good fit if

You have historical AIS positions, port notifications or schedules, and a recurring decision currently based on experience, such as assigning berths, scheduling a lock passage or determining quay capacity.

Not a good fit if

The question comes up once a year, the data is held by a party that won’t share it, or an overview of current use already answers it. In that case, we recommend a dashboard or an agreement.

What determines the scope

How many sources and parties provide data, whether it arrives in real time or periodically, how much historical data is available, and whether the goal is to gain insights or put a model into production.

Is AIS data reliable enough?

Positions are received automatically, but reception depends on location and weather, while voyage details such as the destination are entered on board. That’s why we first check AIS against notifications and schedules.

AIS positions, port notifications or measurement time series: which source should you start with?

AIS positions

Choose this if you want to know where and when ships are moored or sailing, and for how long. We recommend starting here because historical data is usually already available.

Port notifications

Choose this if you want to know which ship, cargo and services are involved in each port visit. In the EU, these notifications go through a national single window.

Measurement time series

Choose this if your question concerns quays, bridges or the water itself. Link the time series to faults and maintenance; see predictive maintenance.

From the port of Amsterdam to the Botlek, we worked with data from the water

Data Science Lab has worked with data and AI since 2016. Based in Amsterdam, we are a Databricks and Microsoft partner. For Port of Amsterdam, Rijkswaterstaat and DutchSail, we worked with data from ships, bridges and sailors.

Book an introductory meeting

In our first conversation, we connect your port data to a single decision

Stef Stoelwinder

Stef Stoelwinder

Business Development

Stef is your first point of contact. In our first conversation, we discuss which port-related question to tackle first and what AIS or planning data you have.

E-mail info@datasciencelab.nl

Phone +31 (0)20 3636 724

What question about berths, waterways or port calls would you like to answer?

Describe your question and the data you already have. You’ll receive an answer by email, or we’ll arrange an initial conversation.

Emailinfo@datasciencelab.nl Phone+31 (0)20 3636 724

Enter a valid email address.

Let us know how we can help you.

Frequently asked questions

AI in ports helps you make decisions that currently rely on experience: how much berth space a type of vessel needs, when a vessel will actually arrive, or when to open a bridge or lock. The starting point is almost always a reconstruction of port calls using AIS data and port notifications.

Often, insight into current use is enough; sometimes a predictive model adds value. This is one of the AI applications we build. To see how data also helps partners in the land-based logistics chain, read about data and AI for mobility and logistics.

AIS (Automatic Identification System) is a system through which vessels automatically broadcast their position, speed and voyage details at regular intervals. According to Rijkswaterstaat, AIS has been mandatory since 2016 on waterways governed by the Binnenvaartpolitiereglement, for commercial vessels from CEMT class I upwards and for recreational vessels longer than 20 metres.

For maritime data analysis, AIS is the most comprehensive source of information on where vessels travel and berth. Reception depends on location and weather, so gaps in the data are normal and need to be handled deliberately when reconstructing voyages.

For inland shipping, it often is. A 2017 study commissioned by the ministerie van Infrastructuur en Milieu points out that, for many skippers, an inland vessel is both a workplace and a home. Unique vessel identifiers and position data can therefore quickly become personal data, and the Autoriteit Persoonsgegevens considers location data sensitive.

For an analysis of port calls, you rarely need data that can be traced back to individual vessels. Define the purpose, retention period and level of aggregation in advance. Your privacy officer or legal adviser can assess whether the GDPR places additional requirements on your application.

Since 15 August 2025, the European Maritime Single Window environment (EMSWe) has applied across the EU under Regulation (EU) 2019/1239. Vessels submit port-call information through a national portal, using a single harmonised dataset and the principle that data is submitted only once per port call.

Under the IMO’s FAL Convention, authorities have been required to use a maritime single window since 1 January 2024. For analysis, this means that notifications of arrival, stay and departure are becoming more comparable and provide a strong second source alongside AIS.

An ETA model learns from historical AIS data how long vessels of a particular type take to travel a route, then adjusts the estimate using their current position and speed. This gives berth or lock planners a realistic arrival time sooner.

The IMO’s Just In Time Arrival Guide describes how a vessel can travel at an appropriate speed and arrive at the pilot station only when a berth, the waterway and nautical services are available. This works only if the port, vessel and service providers share their plans. Read more about optimising planning with AI.

It can support decisions, but the operator or harbour master remains responsible for making them. In the Botlek and Spijkenisser bridges case study, the long-term goal is a system that suggests an opening time to the bridge operator that causes the least overall disruption to shipping, trains and road traffic.

The AI Act classifies AI used as a safety component in critical infrastructure and road traffic as high-risk. Your legal adviser or compliance team can assess whether your application falls into that category. We help identify applications and prepare the risk classification; the timeline is set out on our AI Act page.

Not for every question. For berth policy, historical data updated periodically is enough; for a current arrival time or a live overview for coaches or planners, you need a data stream. For DutchSail, data from the sailing simulator flowed through Kafka to a dashboard in Grafana.

For Port of Amsterdam, we combined real-time data with Databricks and Power BI. The right approach depends on how quickly someone needs to act on the results. A real-time connection requires more management, so we choose it only when necessary; read more about the data and AI platform.

When insight into current use already answers the question. For Port of Amsterdam, the aim was to measure and visualise historical berth use so policymakers could make decisions based on figures. That did not require a predictive model.

A model will not help if AIS data cannot be linked to berths or port notifications, or if partners do not share their data. Start instead with data-sharing agreements or a Power BI dashboard. If you are looking for advice for a specific area, read about AI advice in the Rotterdam region.

What would you like to explore first: the application, the supply chain or the region?

View all services