The Mayflower Autonomous Ship (MAS) is the embodiment of today’s AIpowered, unmanned marine systems. Poignantly named after the merchant ship that transported the first English Puritans to the New World in 1620, this modern-day masterpiece represents a project of transatlantic collaboration like no other, uniting commercial partners Promare and IBM, as well as a raft of other global sponsors.
Following two years of design, construction and training of its AI models, MAS was officially launched on September 16 in Plymouth, UK, fittingly from the very port that bid “fair finds and following seas” to the original Mayflower back in the seventeenth century. However, the MAS project is less about the last 400 years, and more about the next.
September’s unveiling triggered a six-month schedule of sea trials, after which the 15-meter trimaran will attempt a crewless Atlantic crossing, initially slated for April 19, 2021. The hope is that by leveraging the advances in autonomous technology—powered by machine learning and edge computing—the Mayflower Autonomous Ship will prompt a new era for oceanography, typified by the deployment of unmanned systems to safely and efficiently gather critical ocean data pertinent to climate change, micro-plastic pollution, and marine habitats.
AI-INSPIRED DESIGN
Constructed of aluminum and composite materials, and with a weight of just 5 tons, MAS has been designed to optimize vessel stability and hydrodynamic efficiency. Lithium ion-phosphate batteries, in addition to solar panels, power the onboard computer systems and drive the dual 20 kW permanent magnet electric propulsion motors. Top speed is 10 knots, with a payload capacity of 0.7 tons. Autonomous operations are personified by an “AI Captain”; Built by MarineAI and based on a number of IBM technologies, the AI Captain correlates data from MAS’s onboard sensors and systems, including GNSS (Global Navigation Satellite System), IMU (Inertial Measurement Units), radar, weather station, SATCOM, and AIS.
MACHINE LEARNING (ML) AND EDGE COMPUTING
Mayflower technicians have used over a million nautical images to finetune the ship’s AI models—and training is ongoing. To meet ML processing demands, the team relied on an IBM Power AC922 fueled by IBM Power9 CPUs and NVIDIA V100 Tensor Core GPUs, the same technologies behind the world’s smartest AI supercomputers. This advanced computing system enables autonomous object detection, classification, and avoidance. While at sea, MAS will process data locally, increasing the speed of decision making and reducing the amount of data flow and storage on the ship.
BEYOND OCEANOGRAPHY
The implications of a successful and unimpeded transatlantic crossing are far-reaching; in many ways, the Mayflower is as much about autonomous marine operations as it is oceanography. Brett Phaneuf, Cofounder and Co-Director of the Autonomous Ship project, told ON&T:
“Autonomy clearly is key to the development of unmanned ships conducting research and commercial operations in the future. But perhaps the greater, more immediate need for autonomy is on manned vessels of today. This may sound counter-intuitive, but there is an important role for autonomous systems on the manned ships, acting as a co-captain, maintaining situational awareness, providing recommendations and decision support to the human crew. Think of it as ‘Augmented Intelligence’: machines helping people with the efficient and safe navigation of the ship.”
The extent to which autonomy could transform the maritime industry remains to be seen, but pioneering initiatives like MAS will prove fundamental to refining options and extending our technical capabilities.
For more information: www.mas400.com
This story was originally published in ON&T’s November 2020 issue. Click here to read more.
