intelligent automation:
teaching ai to sail


Modern ocean racing sailing boats are high performance machines, almost more comparable to aircraft than the yachts of old. They combine cutting edge material science, aero and hydrodynamics, navigation systems, telecommunications, and sensors.

IOT data for performance optimisation and intelligent automation

Modern racing boats have fully integrated internet of things (IoT) with sensors ranging from wind speed and direction, to load meters, autopilot actuator sensors, accelerometers, GPS, echo sounders (depth), water logs (speed through water), and more. This generates a large and diverse set of data.

One underdeveloped technological domain is the use of real-time data from boat sensors for automated performance optimisation. While data is relayed via displays to a human, there is little or no interfacing with the autopilots that are vital for long distance racing with only one or two crew.

The performance gap

Autopilots for sailing boats are comparatively crude. They are responsible for >95% of steering on single handed sailing races. However, they only achieve 80% of the performance of a human. Statistically just a 2% performance increase on the previous edition is all that is required to win the Vendee Globe.

There is an enormous potential to provide a race winning advantage through the development of machine learning driven AI for these autopilots.

A high performance test platform: Concise 8, Class 40

Concise 8 was developed in 2013 as a third generation Class 40. Designed at the cutting edge of technology utilising techniques developed during America’s cup, Volvo Ocean Race and IMOCA 60 campaigns, she is a sleek, high-speed, state of the art machine.

  • pencil-and-ruler-cross-of-school-materials

    Dimensions – 12.19m long, 4.5m wide, 3m deep and 19m tall

  • food-scale-tool

    Weight – 4500kg all up, with over 2000kg in the keel

  • area

    Sail Area – With upwind sail area limited to 115m2, Concise 8 sports a massive 250m2 downwind

Whilst newer boats have been launched since the inception of Concise 8, few have demonstrated the speed and capability of her radical design, so she still remains a strong favourite in all races she enters.

Working alongside Jack Trigger Racing and control and sensor systems manufacturer NKE, we are harvesting highly granular IoT sensor data directly from the boat’s processor.

the human sailor, jack trigger

Jack is the rising star of british offshore sailing. He is ultimately driven to compete in the Vendee Globe and challenge to become the next successful British skipper in the race. His dream is to follow in the footsteps of Dame Ellen McArthur and Alex Thompson, and even to go one better than Alex who came second in the last edition.

Competing in the Class 40 division with maximum support and achieving good results in the championship and Route du Rhum are a major step towards achieving this.

Age – 25yrs
Height – 6ft
Nationality – British

Previous results:

  • 8th of 53 Route du Rhum solo transatlantic race
  • Line honours Rolex Fastnet Race 2017 – MOD 70 Concise 10
  • 2nd Rolex Fastnet Race 2015 – Skipper Concise 8
  • 1st RORC Season Championship 2014 + 2015 – Skipper Concise 8
  • 2 Transatlantic’s and RORC Champions – MOD 70 Concise 10
  • 3rd Line honours Rolex Middle Sea Race 2017 – Hugo Boss
  • Part of record breaking Artemis crew in Length of Britain Challenge 2017

With an engineering background from Oxford, he is the ideal sailor to work with us on the development of an intelligent autopilot solution. He is also a type 1 diabetic, a condition that is problematic for an extreme sportsman. Admirably, Jack is at the forefront of a technological push to help manage the condition, as well as being a role model for young people that share it with him.

our goal: the vendee globe

AI at the extreme

The Everest of the Seas

The Vendee Globe is the only sailing race around the world, non-stop, and without assistance. Known as the ‘Everest of the Seas’, only 167 competitors have taken part, with 89 completing the race and 6 winners.

While competing, the sailors are more isolated than astronauts on the ISS, both in terms of distance from land, solitude, and support (they have none). Connection to the outside world is only via satellite link.

Formula One of the Oceans

The modern boats for this race are truly at the cutting edge. Built of lightweight carbon fibre, and now with hydrofoils, they are punishing to race for the human sailor as they travel across the sea at up to 65 kilometres per hour. It is in this context that autopilots are used to steer the boat the vast majority of the time (up to 95%), while the human sailor adjusts the sails, navigates, and manages the system as a whole. The sailors only sleep 20 minutes at a time for over 70 days.

It is into this environment that we want to place AI, test it at the extreme, and provide a race winning edge.

A Technology Team Up to the challenge

The project involves a number of partners from across academia and industry: Jack Trigger Racing, Microsoft, Imperial College London Aeronautics division, and WisCont. In collaboration with our partners, T-DAB is aiming to develop cutting edge machine learning driven artificial intelligence to deploy to standard, off the shelf autopilot hardware.

The Data Analysis Bureau is an end-to-end data science and advanced analytics service provider that believes "smarter data means better decisions".

Jack is ultimately driven to compete in the Vendee Globe and challenge to become the next successful British skipper in the race.

We regularly collaborate with WisConT, our specialist technical partners, with high-end expertise in data, industrial IoT, and infrastructure.

We are partnering with students from the Imperial College, which is a world-class university with a mission to benefit society through science.

We are a member of the Microsoft Partner Network, working towards becoming a key specialist ecosystem partner.


Combining ML/AI, IOT and Edge computing
powered by Microsoft

The project provides an ideal opportunity to test, develop, and show case the use of the latest machine learning, IOT, edge computing and Cloud solution technologies.

In particular, the team plan to use Microsoft technologies that allow for training of algorithms in the cloud, streaming of data from remote connections, and deployment of algorithms to remote, low power computing hardware.


IOT data for performance optimisation and intelligent automation

To bring more intelligent automation and control to marine autopilots, we are developing, testing, validating and benchmarking a number of machine learning models. These include both deep reinforcement learning and advanced recurrent neural networks (RNNs).

In order to do this effectively for the amount of data required, we a supported and powered by Microsoft and their cloud based Azure Databricks solution (hyperlink). Azure Databricks is optimised for deep learning, supports multiple languages and libraries, integrates with other Azure services, and provides an ideal collaborative working space.

AI at the Edge, powered by Microsoft Azure IoT Edge

Ocean racing sailing boats are extreme environments for technology. They are isolated, connected only by satellite links with limited and expensive bandwidth. In addition, power onboard is at a premium with the priority being weight reduction for speed. This means that computing power is limited, presenting a challenge for the deployment of machine learning algorithms. Nor can they be connected continuously to the cloud.

Microsoft Azure IoT Edge provides the ideal solution to solve this problem. It allows for the storage of data and development and training of machine learning algorithms in the Azure cloud, while then optimising the deployment of these algorithms to remote, low power computing as close as possible to the point of action.

There is no more suited use case for this technology than a sailing boat in the middle of the Southern Ocean!

Find out more about why we are using Azure IoT Edge...

Deploy ai and analytics to the edge

IoT Edge allows us to deploy complex machine learning and other high-value artificial intelligence to the autopilot processor which is an edge device.

Reduce iot
solution costs

Only a small fraction of the boat’s IoT data acquired is meaningful post-analytics and model training. Using trained models to process the data locally and send only what’s needed to the cloud for further analysis or benchmarking. This reduces the cost associated with sending all of the boat’s data to the cloud while keeping data quality high.


IoT Edge holds to the same programming model as other Azure IoT services; for example, the same code can be run on a device or in the cloud. This means our data scientists can work to improve the machine learning behind the autopilot AI, only deploying it locally to the boat when ready.

Operate offline or with connectivity

With IoT Edge, the autopilot (edge device) will operate reliably and securely even when it is offline or has intermittent connectivity to the cloud. Azure IoT device management automatically syncs the latest state of the pilot once it has reconnected.

Project Team

Eric Topham

CEO and Data Science Director

Ivan Scattergood

Data Engineering Director

Pedro Baiz

Associate Researcher

Roman Kastusik


Birk Ulstad


Stay up to date with the project

February 10, 2020

The opportunity cost of Machine Learning

What is the opportunity cost of machine learning? An opportunity cost can be defined as…

Read more
January 21, 2020

Top 5 manufacturing issues solved by ML

The manufacturing industry faces continual disruption from changing markets, new product demands and technological innovations. Whether it’s known as Industry 4.0, Smart Factory or…

Read more
January 9, 2020

The ML & AI Opportunity in 2020

The AI & ML opportunity in 2020 for many companies starts by simply by observing how other businesses…

Read more