What does the autonomous ship, Mayflower 400, have to do with business-driven decision-making? Quite a bit, it turns out. As you’ll see, the development of this amazing, one-of-a-kind vessel led to a conclusion that we at Decision Management Solutions see every day in our client work: It’s never enough to just rely on artificial intelligence (AI)/machine learning (ML) to do all the decision-making. On the contrary, optimal decision-making is based on the marriage of artificial intelligence technologies and business rules (created by humans)—and this is what makes real-world scenarios possible.
Are you ready to dive in to learn more about the evolution Mayflower 400? I think you’ll see the parallels with Decision Management as the story unfolds.
Marine AI—based in Plymouth, U.K.—in partnership with IBM and other organizations is in the process of conducting performance tests on the world’s first autonomous ship and is due to set sail for its first transatlantic voyage this spring. Early on, the project started as a grassroots effort by Marine AI, with the collaboration of smaller ocean industry companies. But, as the project progressed and IBM AI, cloud, and edge technologies were being integrated into the system, IBM got more deeply involved.
The goal is to establish the Mayflower 400 as an open platform for marine research that would reduce costs and ease the burden on scientists and sailors, who have to brave a dangerous and unpredictable environment in the course of their data-collecting missions. Some of oceanographic scientific experiments that will be conducted by Mayflower 400 in the foreseeable future include measuring temperatures, taking water samples to test for microplastics (via IBM Hypertaste, an AI chemical sensing system that acts as an “ocean tongue” that classifies the digital footprint of liquids) and—perhaps most intriguing of all—identifying whale acoustics to track and study these immense ocean-dwelling mammals.
The futuristic Mayflower 400 is completely uncrewed and travels at 8 knots (~9 miles per hour), with a burst speed of up to 16 knots (~18 miles per hour). As a completely autonomous entity, all the ship’s decision-making ability is resident on the vehicle itself. To achieve this required a confluence of many technologies and capabilities—from hazard detection by radar to GPS for navigation to a vehicle management system.
Here’s how Rob High, VP and CTO for edge computing at IBM, describes the project: “This is a test bed for integration of AI/ML and deterministic rule-based decision-making. The Mayflower is an edge device, with a computer on board and a low-bandwidth connection to shore, so all decisions are made onboard. There are symbiotic relationships that enable the ship to operate independently at sea: training the AI model, systems integration, and operational COLREGs.”
COLREGs or the International Regulations for Preventing Collisions at Sea, were formulated by the International Maritime Organization to prevent seagoing vessels from colliding. These are essentially the “rules of the road” at sea. For the Mayflower 400, a critical part of understanding its environment is knowing how other ships behave and how they comply with COLREGs rules. This is a key element in the safe operation of the vessel itself. The “AI Captain” onboard the Mayflower 400 will draw on IBM’s business rule management system, Operational Decision Manager (ODM), to ensure the vessel is following COLREGs. The great thing about leveraging ODM in this context is that it provides a completely transparent record of its decision-making process. These abilities—accurately implementing regulations and best practices, as well as creating a transparent record of how decisions were made—are similarly critical drivers of our use of business rules in Decision Management.
Let’s turn to the AI technology part. An essential capability for the Mayflower 400 is understanding what’s going on around it in its test environment—the busy waters at the port at Plymouth Sound, which is completely outfitted with an advanced technology infrastructure. One of the Mayflower 400 team’s first undertakings was to develop computer vision models based on different types of artificial intelligence—namely machine learning and deep learning. The engineering team fused data from multiple sensors, along with information from weather surveillance radar, to develop a hazard chart. The AI Captain on the ship learns how to determine an optimal and safe path in a complex environment and is able to make the right navigation decisions. Since the ocean is a dynamic and sometimes hostile environment, communication with a shore-side system is not always possible. Ultimately, the ship needs to be able to make smart decisions all on its own, so the intelligence needs to be resident onboard.
As High points out: “It’s never enough simply to use ML by itself to do all the decision-making processes… This is a combination of both traditional deterministic-rules decision-making with AI-based imprinting. And these two things together make these real-world scenarios possible. That’s kind of what humans do—we are highly informed by past experiences and our own set of rules informed by those experiences.” I couldn’t have said it better myself.
Don Scott, CTO and chief project engineer, elaborates, noting that, ultimately, Mayflower 400 depends on AI combined with human-generated rules. “The AI Captain is a distributed system—not just one monolithic application. It’s made up of a series of different systems operating and interacting with each other. Ultimately, it’s about providing a safe action and then doing it—which is the role of a human captain,” he says. “We’re finding that it doesn’t have to be completely autonomous. It’s really augmented intelligence—sharing the cognitive load between human and machine—and having these systems assist humans in their everyday tasks and improve the safe operation of their vehicle.”
It’s apparent that ensuring the Mayflower 400 is capable of making the right decisions at the right time and under the right circumstances will be crucial for the success of its maiden voyage. Similarly, Decision Management must focus first on the business decision and use the combined power of AI, data, and business rules to achieve positive and desirable business outcomes.
Click here to learn more about our approach to integrating business rules-based decision-making with advanced machine learning and AI.
If you’d like to view the webcast replay of “To the Edge with the Mayflower Autonomous Ship,” visit:
Watch the 8-part docuseries from IBM that provides an inside look into the creation and first-time voyage of the Mayflower 400.