This autonomous ship aims to steer itself across the Atlantic ocean

The Mayflower Autonomous Ship will use IBM’s servers, AI, cloud and edge technologies to travel from Plymouth, UK, to Plymouth, USA.

Olli: The self-driving bus with IBM Watson onboard A new Local Motors autonomous vehicle is using the cognitive computing system to chat to passengers.

An autonomous boat under developments could be the first ship to cross the Atlantic that is able to navigate around ships and other hazards by itself.

The Mayflower Autonomous Ship (MAS) is an autonomous vessel due to depart from Plymouth in England on the fourth centenary of the original Mayflower voyage, on 6 September 2020, with its destination Plymouth, USA. 

The project was put together by marine research and exploration company ProMare in an effort to expand the scope of marine research. The boat will carry three research pods equipped with scientific instruments to measure various phenomena such as ocean plastics, mammal behaviour or sea level changes.  

IBM has now joined the initiative, and it will supply technical support for all navigation operations. 

mayflower1-jpg-web-e1567152206926.jpg

The Mayflower Autonomous Ship (MAS) is an unmanned vessel set to depart from Plymouth in England on the fourth centenary of the original Mayflower voyage. 

Image: ProMare

To avoid obstacles, the ship will be loaded with various sensors - similar to those used for autonomous vehicles - including radar and Lidar as well as automatic identification systems (AIS), which is the conventional tracking system vessels use to exchange their GPS position. 

This will be combined with IBM's Power AI Vision - a data processing technology that blends computer vision and deep learning to identify and track objects both static and moving.

Andy Stanford-Clark, chief technology officer at IBM UK, told ZDNet that the system has been trained for two years on the different types of obstacles in could encounter. 

"Anything solid will cause a reaction from one of the sensors," he said. "And because we have fused various different sensors, we are pretty confident that we will consistently know what's in front of us."

When objects are detected, MAS will use IBM's operational decision manager (ODM) software, which takes the input and combines it with the rules of navigation that it was taught to decide whether to change its course. 

"ODM will navigate around obstacles but also keep track of the overall mission," said Stanford-Clark. "For example it will remain aware of weather data to avoid dangerous routes and determine the optimal path. It's a rules-based system that lets it make the right decision."  

Autonomous vessels are a hot topic in the shipping industry, and IBM is not the first tech giant to look into AI-powered naval crews. 

In 2017, for instance, Rolls-Royce announced that it was working with Google to train its AI-based object classification system and improve situational awareness for ships. A year later, it teamed up with Intel to create designs for intelligent shipping systems.  

But if MAS is successful next year, it will be the first ship to complete the ambitious trans-Atlantic trip via a route that was not pre-programmed.   

Throughout the journey, data collected will be stored locally and then uploaded to IBM Cloud when satellite connectivity is available. This includes data for maritime research, which is the primary purpose of the mission; but it will also provide operators with feedback on the ship's own performance.

Stanford-Clark said that MAS will be a research project in itself, and there is still risk in sending it across the Atlantic; "we cannot be 100% certain it will work," he said. In any case, though, it will provide the team with useful information to improve on the technology.