Pushing the limits of autonomous navigation with AI and automation
Groundbreaking technology from a new, fully self-operating ship inspires the next generation of innovation for your business
Get the latest updates on the Mayflower's voyage
Overview
A Mayflower to advance the promise of AI and automation
The culmination of a multi-year project with IBM and marine research non-profit ProMare
The maritime research vessel Mayflower Autonomous Ship is a first-of its-kind autonomous ship — and IBM technology played a central role in bringing it to life. On its inaugural journey, the ship will commemorate the original Mayflower by following its transatlantic route. But this modern Mayflower will gather critical ocean data on the impact of climate change and pollution so that marine researchers can better understand and protect our oceans — now and into the future.
IBM and ProMare co-engineered the software for the crewless ship with three layers of technology: sensory inputs, real-time machine learning and analytics, and a decision engine. IBM and ProMare experts used petabytes of data to train machine learning models and wrote rules-based decisioning for the decision engine, enabling the ship to react to an often-treacherous ocean environment — with zero human intervention. It adheres to maritime law while making crucial split-second decisions. It reroutes itself around harsh weather environments. It collects and analyzes massive amounts of ocean data. And it does it all 24/7.
By extending the ship’s groundbreaking intelligent automation, operational decision-making, edge computing and AI-powered remote monitoring technologies across industries, we can pave the way for the next generation of innovation, efficiency, safety and cost-savings in your business.
Read the IDC paper (link resides outside IBM) (PDF, 466 KB) →
Miles without onboard crew:
3,349
Onboard sensors gathering data:
30
Edge devices analyzing data:
15
AI Captain
Launching a crewless ship requires technology that can power autonomous intelligence. IBM built the cognitive pilot that makes all of the ship’s decisions, nicknamed the "AI Captain."
The new Mayflower is a huge step forward in building a system that has opinions about how to act and can then take decisive action — all on its own.
The decisions that the AI Captain makes are based on if/then rules and machine learning models for pattern recognition, but it’s more than just the sum of those parts. It learns from the outcomes of its decisions, makes predictions about the future, manages risks and refines its knowledge through experience. And it does it by integrating more inputs in real time than a human can do alone.
For large portions of the inaugural trip, there won’t be the ability for a human to intervene. The ship will need to make the best decisions on its own. We’re creating similar self-operating technologies for businesses.
See how you can give your business its own AI Captain (PDF, 941 KB).
Boost performance across operations
Intelligent decisioning used by the AI Captain is just one piece of a larger AI-powered automation solution for business operations. Businesses also rely on IBM automation for tasks, workflows, content services and process mining — capabilities that come as part of a set of integrated software designed to help solve even the toughest operational challenges.
Operate with smart decisioning
Where the AI Captain runs sensory data and AI-annotated models of the surrounding context through IBM’s automated rule management software to modify routes and avoid collisions, businesses already use the software for the decision-making records it provides when, for example, approving loans and personalizing customer offers.
Act on insight at the edge
The AI Captain can make decisions because 15 edge devices, orchestrated by IBM software, process data from 30 sensors, including radar, GPS, cameras, and altitude and water-depth detectors. With the same software, businesses can scale and run edge solutions anywhere, acting on insight closer to where data is created.
If we compare the components of the AI Captain to a human…it has to recognize, build a plan and then act on that plan.
Andy Stanford-Clark
Chief Technology Officer
IBM UK/Ireland
How the ship works
Sensory inputs
Sensory inputs
The dozens of sensory devices on the ship include radar, GPS, weather, and altitude and water-depth detectors as well as high-definition onboard cameras that send visual input to an IBM computer vision system. The system has been trained to identify potential dangers, including cargo ships, fishing vessels and even partially submerged shipping containers floating in the water.
Real-time analytics
Real-time analytics
Using inference algorithms and models generated from IBM computer vision technology, the ship was trained on more than one million nautical images to recognize other ships, debris, bridges and other hazards. ODM evaluates international collision-preventing regulations regarding nearby vessels, generates a risk map indicating “unsafe” situations and makes recommendations.
Decision engine
Decision engine
The AI Captain ingests ODM recommendations, interprets computer vision inputs and other data, and analyzes forecasted weather to determine how to avoid hazards. Additional Marine AI math modeling provides decision support on the best action, including instructing the ship to change course or speed. The ship’s Safety Manager verifies decisions are safe, even allowing the AI Captain to make split-second decisions.
