USVs Fuse Intelligence, AI Automation, and Agility to Progress Ocean Research

As human activity continues to impact the health of our planet, it is vital we harness the power of modern science and technology to safeguard our ocean ecosystems, find new resources, and build futureproof societies.

Uncrewed surface vehicles (USVs) like the Mayflower Autonomous Ship (MAS) are fusing intelligence, automation, and agility to help chart a more sustainable relationship with the ocean.

MAS uses cameras, RADAR, AIS and other sensors to detect and avoid obstacles. (Image credit: IBM/ProMare)

A New Transatantic Journey

Guided by a new “AI Captain” and built using IBM cloud, Artificial Intelligence (AI), and edge computing technologies, MAS will re-embark on a three-week transatlantic journey in April 2022. This is our second attempt at a transatlantic crossing. The first attempt in 2021 was thwarted by a mechanical issue when a coupling broke on the hybrid generator exhaust system, forcing us to abort the mission. After a few months of repairs—essentially making a new custom exhaust and reworking how it fit into the ship—we began a new series of sea trials, in December 2021.

This initial setback was difficult for the entire team. But the only way to achieve true autonomous performance with any level of predictability is to suffer through the unpredictable parts now. Innovation is never achieved without risk; and by comparison to our namesake, the original Mayflower that crossed the Atlantic in 1620, our risk is infinitesimally low.

Applying Intelligence, AI Automation, And Agility

Navigating an ever-changing seascape depends on the ability to collect and process oceanographic data locally so that mission-critical decisions can be made. To date, this has somewhat limited the applications for uncrewed vessels. MAS is making tangible progress to overcome this challenge.

With no human assistance for the entirety of the voyage, MAS will use cameras, RADAR, AIS and other sensors to intelligently detect and navigate around obstacles, solar panels to generate green energy, and an operational decision manager to weigh marine regulations and protocols (COLREGS) against environmental conditions and vessel operating status. Computer vision technology deciphers images streaming in from the ship’s six on-board video cameras. Trained on millions of maritime images gathered since 2016, this vigilant AI-powered system is now able to recognize a wide range of threats in the vicinity of the ship—whether physical outcroppings of land, floating debris, marine life, other vessels, or various other potential perils. Even paddle boarders!

The AI Captain technology we’ve installed was created using agile methodologies. Even before the ship was built, the AI Captain machine learning models were built and tested on an IBM AC922 Power server—the same technology behind some of the world’s most powerful supercomputers. This reduced the time and expense of physical testing and means that MAS’s algorithms can be constantly tweaked and updated by the MAS support team on shore. It also means that the AI Captain can be easily installed onto other vessels. 

To facilitate ocean research, MAS has 3 payload bays for scientific equipment and the capacity to carry 1,000 kg of scientific equipment. Research pods located on MAS collect data relating to a multitude of inputs around ocean health and chemistry: sampling water for microplastics, monitoring sound patterns associated with marine life, and measuring water depth and determining sea surface heights and wave patterns. When MAS achieves connectivity, it will upload this data to the cloud for researchers from IBM, ProMare, MarineAI, University of Plymouth, Newcastle University, University of Exeter, and University of Liverpool to access. Live weather data will be downloaded from The Weather Company to optimize performance of the AI Captain. When network connectivity isn’t available, MAS’s on-board weather station keeps the weather data flowing. The AI captain fuses different data sets together to make rapid and informed decisions.

AI at Sea

In the world of finance, retail and commerce, automated decision management systems are used to analyze customer activity and automate the process of recommending relevant offers and services. Such automation also enhances privacy and security, identifying possible cases of fraud, as well as reducing the amount of personal and sensitive information that staff members are required to process. 

Researchers from IBM, ProMare, MarineAI, and several UK universities will have direct access to real-time ocean data during the uncrewed voyage. (Photo credit: IBM/ProMare)

Likewise, MAS needs to be able to operate independently through its own decision automation system. This Operational Decision Manager (ODM) system assesses all available data against a pre-determined set of rules (COLREGS, among others) to help the AI Captain make the best decision in response to real-time events. ODM accesses a broad range of data sources, including the ship’s computer vision system, weather data, radar, sonar, and other marine navigation systems to better understand the surrounding environment. And because it is trained on two key sets of rules—International Regulations for Preventing Collisions at Sea, as well as International Convention for the Safety of Life at Sea—ODM helps MAS follow maritime regulations, while considering real-time data to optimize its decision making.

MAS, like many businesses, also has a decentralized computing architecture. Edge computing is essential for enabling the autonomy required for continuous operations. Small, powerful, and lightweight edge devices provide just enough compute power for an uncrewed vessel to operate independently, even without network connectivity. When a connection with the network cloud is made, performance data is uploaded, system updates are downloaded, and the autonomous system continues. 

MAS has agility built into its core. Its trimaran hull form ensures high levels of hydro- and aero-dynamics. Waterproof photovoltaic panels fitted to its upper surface enable it to draw on energy from the sun when available, with a modern, high-efficiency generator acting as backup in case solar power is low in supply.  

The ship has high levels of redundancy built into it, meaning that its systems are both isolated and duplicated to reduce to chance of a single-point failure. And rather than mindlessly following a pre-charted course, MAS’s AI Captain handily evaluates all available data and constantly updates the ship’s route and speed, second by second. 

As we continue to explore extreme ocean environments, our vision is for humans and machines to continue to work in tandem, using agile, intelligent, and autonomous solutions to augment and extend people’s capabilities. It is smart tech at its finest.

For more information, visit: www.mas400.com.

This feature originally appeared in Ocean News & Technology's Magazine January 2022 edition—Uncrewed Surface Vehicles (USVs). To read more, access the magazine here.