IBM and ocean research non-profit ProMare have completed and launched the Mayflower Autonomous Ship (MAS), an AI and solar-powered marine research vessel that will traverse the oceans gathering important environmental data. Its main test and voyage will be to travel autonomously from Plymouth, England, to Plymouth, Massachusetts, much like the original Mayflower, which sailed from Plymouth Sound to the New World in 1620.
The main purpose of the project is to test and validate AI on the edge to drive development of full marine autonomy. The broader implication and residual impact is learning how to enable system-wide automation of a new sector, from manufacturing and sourcing through to the endpoint itself. While fully autonomous commercial vessels will not be in the market any time soon, the initiative is an important proof of concept with key takeaways for what it takes to stand up an autonomous solution, plus the potential transferability of technologies across end markets.
Under Way, Making Way
|
IMPACT
|
There are four main technology components for automation:
- AI: MAS has been developed to navigate independently without human intervention, which means following the rules of maritime traffic (72 COLREGS) in addition to events-based obstacle recognition and avoidance. To do this, the vessel must have its own compute resource and operate independent of any terrestrial command and control. This new class of marine AI is underpinned by IBM’s advanced edge computing systems, automation software, computer vision technology, and Red Hat open source software. The 15 meter-long MAS achieves an autonomy level of 5 (the equivalent of a driverless vehicle) with more than 30 onboard sensors, 6 AI cameras, and a slew of software that includes IBM Visual Insights computer vision technology, IBM edge systems, IBM Operational Decision Manager automation software, IBM Maximo asset management software, and data from The Weather Company.
- Computer Vision: Computer vision is integral to environmental sensing and navigation. MAS uses Maximo Visual Inspection, which provides a sandbox for deep learning computer vision model development, and IBM Edge Application Manager for the deployment of those models onto the ship. This same thinking is core to enabling higher degrees of automation on the factory floor.
- Edge: All of the ability for the MAS to understand and make decisions is resident on the ship itself. Interestingly, the project is using IBM Operational Decision Manager for certain aspects. IBM Operational Decision Manager is a deterministic rules engine often used in financial services, signaling the transferability of certain products/capabilities used in automating other industries to maritime (and perhaps, in turn, manufacturing). In terms of hardware, the vessel employs IBM Power Systems AC922, 6 Jetson AGX Xavier, 2 Jetson Xavier NX, 4+ Intel-based computers, and 4+ custom microprocessor systems.
- Connectivity: Unlike on land, oceangoing vessels cannot rely on the typical connectivity suite past the demarcation line that divides inland and international waters. Vodafone is working with IBM to allay some of these concerns by providing 5G connectivity to the MAS in initial tests taking place in Plymouth Sound. However, ABI Research believes that intercontinental/transatlantic 5G is an unrealistic expectation for anytime in the foreseeable future. A more likely scenario is to employ 5G at ports and select intracoastal waterways in the near term, relying on satellite constellations like SpaceX’s Starlink service once the network is commercially available and tuned for such operations.
Launch and Go
|
RECOMMENDATIONS
|
The international shipping industry is responsible for the carriage of around 90% of world trade. Containerships, for example, have the capacity to carry several large warehouses worth of goods on a single journey, and car carrier ships can handle roughly 7,600 cars per instance. Without shipping, the bulk transport of raw materials, intercontinental commerce, and the import/export of affordable food and manufactured goods would not be possible. Furthermore, 71% of the world is water. This means untapped opportunities for greater autonomy in not only freight transport but also pleasure boating, environmental and scientific research, and short- and long-range transit ferries.
Ships are technically sophisticated assets that cost tens or hundreds of millions of dollars to commission. On a smaller scale, the high price of pleasure boats makes such an activity unattainable for many. Core to this high cost is not only the cost of materials but also labor and, more important, the processes knit together to bring a concept to life—many boat and ship building operations remain highly manual/analog. Maritime manufacturing, much like the maritime industry, is far from automated.
The successful completion of the MAS’s maiden voyage from Plymouth, England, to Plymouth, Massachusetts signifies a turning of the tides whereby boat builders and industry participants will begin to open their eyes to the benefits and potential of digital transformation. Automation of the endpoint (vessels in this case) means embedding novel technologies like 5G, edge computing, machine vision, and AI. We’ve seen the knock-on effect of this new thinking in the automotive market, which is undergoing a panacea of change with respect to electrification and the processes employed to manufacture a new breed of offering; automotive remains one of the most automated industries aside from semiconductor fabrication. It will be several years before the market sees the direct impact and result of the MAS, but the initiative and involvement of companies like IBM, Red Hat, and Vodafone demonstrate a direction of travel that cannot be ignored.
