Register for a free trial

Maritime Digitalisation & Communications

Here come the machines

Mon 06 Aug 2018 by Selwyn Parker

Here come the machines
Artifical intellegence relies on serious computer power. Server room at BP Centre for High Performance Computing (credit: BP)

The maritime sector is embarking on an era built on artificial intelligence

Norway-based fleet operator Wilson ASA has embarked on a probably endless voyage with artificial intelligence (AI). Over the coming years the group expects to improve navigation, communication, connectivity and fleet data among other vital elements involved in running its 100-strong fleet.

In early July, Wilson signed one of the first big AI-driven contracts in the maritime sector to apply machine-learning techniques on ship and shore. Working with AI specialist, the Wartsila-owned Transas group, Wilson’s so-called fleet operations solution enables a highly collaborative environment between its bulk carriers and the offices on shore.

“Digital solutions can add value to large fleet operations”

The plan is to connect fleet operations with the ships, port and coastal traffic management through sharing mutually beneficial data. Wilson’s goal is exactly what most fleet operators fervently want – better safety, reliability, efficiency and higher profits in a competitive world.

Thus, with AI being taken to sea, the digital world is becoming more and more embedded in the maritime sector. “Digital solutions can add value to large fleet operations,” explained Transas leader for Wartsila Voyage Solutions, Frank Coles. “Joined-up technology has a definite role to play in the sector’s transition to a new era of efficiency.”


As the realisation dawns that the constantly mounting complexities of shipping present challenges beyond the ability of human intelligence to provide perfect – or nearly perfect – solutions, other companies are jumping on the AI bandwagon. After all, the thousands of cargo ships that ply the oceans constantly run into difficulties, unforeseen or otherwise. Hostile weather, congested ports where vessels are forced to wait at anchor, equipment failures, full containers on the way out and half-empty ones on the outward voyage, manning problems: they can add up to a logistic nightmare where operators, however experienced, could do with help in the shape of AI.


One of the latest to recognise the promise of AI is Orient Overseas Container Line (OOCL) which has a track record of embracing technologies that fall straight to the bottom line. So committed is the Hong Kong-headquartered group, with a fleet of nearly 60 vessels, it employs a large team of developers around the world whose job is to create software that makes the skipper’s job easier.

Earlier this year, OOCL engaged a machine-learning consultancy, Microsoft Research Asia, to map out its digital journey. After spending 15 weeks analysing ways to improve its shipping network operations, the group got immediate results.

According to OOCL’s chief information officer, Steve Siu, the result of that collaboration slashed about US$10M a year from operating costs. With economies of that scale lying on the table, the shipping company promptly took a big step further in April by signing an 18-month partnership to embed two fundamental elements of AI – deep learning and reinforcement learning – into its operations.

The vagaries of running a big fleet of ships seems tailor-made for what machine learning can offer. “Shipping network operations involve multiple parties and variables that can change at any moment,” explained Microsoft Research Asia managing director, Hsiao-Wuen Hon.

Having identified the long-term opportunities, the partners are throwing a lot of resources into AI. Right now, Microsoft Research Asia is training more than 200 machine-learning developers. (There is a global shortage of AI specialists.)


But what exactly is artificial intelligence, also known as machine learning? According to the Oxford English Dictionary, it is “the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision making and translation between languages.” And as we see further down, most of these tasks have direct application to the maritime sector, including ports. Land-based manufacturers were much quicker to spot the value of AI and implement its techniques in what is known as Industry 4.0.

Global consultancy McKinsey is a fervent believer in the potential of AI, while recognising the challenges of installing it in a real-life situation. In a study entitled Crossing the frontier: how to apply AI for impact, McKinsey analysed 40 cases where the application of machine learning beat – or enhanced – human intelligence in resolving complex issues.

To take just one example, that of predictive maintenance, increasingly seen as fundamental in the post-2020 shipping world, the consultancy said “Deep learning’s capacity to analyse very large amounts of high-dimensional data can take existing preventive maintenance systems to a new level.”

Data-driven predictive maintenance is particularly valuable in the case of hard-worked assets such as engines. Although the following example is not a nautical one, it is relevant to shipping. McKinsey found that AI “can extend the life of a cargo plane beyond what is possible using traditional analytic techniques by combining plane model data, maintenance history, internet of things sensor data such as anomaly detection on engine vibration information, and images and video of engine condition.”

Nor does all this necessarily cost the earth – microphones and cameras can capture a lot of predictive data.

Virtual twin

The ship construction industry might be ahead of the vessel operators. The former has already discovered AI in the guise of the digital – or virtual – twin. Effectively a phantom version of the real thing, the digital twin mirrors the physical structure. Currently, more and more complex vessels such as floating production and storage and offloading platforms (FPSOs or floaters) are being developed with a broad range of digital tools that create a working model in 3D.

Although the digital twin is virtual, the benefits are real enough. A giant FPSO, with a capacity of 220,000 barrels a day, was launched in mid-2017 after the construction job was shortened by nine months, saving US$15M.

In another example of AI adding value to these complex projects, the crew of FPSOs and other vessels can be trained in simulated work situations. Through a Siemens-developed programme called Walk Inside that creates real-life scenarios, crews became familiar with their tasks long before they get down to the business of pumping oil.

And throughout the construction process, all the parties involved – designers, engineers, IT experts and owners – are kept in the loop through 3D visuals. Rather than having to pore over printed manuals, they can see on screen everything that matters, from the integrity of the welding process to the cabling.

Topsides 4.0

The Siemens group has developed a package called Topsides 4.0 that is slashing costs for constructing big and complex vessels. It also cuts costs when the vessel is up and running – Siemens told the author that operating expenses for a mid-sized FPSO over a 10-year period will fall by more than US$100M by applying programme, because the package harnesses the streams of data that sensors pour out day and night.

Known as automated intelligence monitoring, the data is streamed from vital assets, especially the so-called rotating ones such as pumps and heat exchangers. Customised analytics are assigned to each of these assets so they can be packaged and analysed relatively simply, albeit by experts.

As Siemens’ director of digital solutions, Elgonda LaGrange, said “[Topsides 4.0 came about] because we saw that we had a portfolio of intellectual knowledge that we could assemble into an overall solution that would make a dramatic difference in the oil and gas industry.” And surely, in the shipping industry.

Siemens had already seen the potential of its package first hand in the automotive sector, one of the first to embrace AI on the assembly line. When Fiat Chrysler engaged Siemens to install machine learning in its Maserati plant in Modena, Italy, production jumped by 30%.

As oil and gas consultancy Wood Group explained “All offshore oil and gas businesses are being prompted by the internet of things [an offshoot of machine learning] to shift their operational paradigms into the digital world. FPSO contractors and producers are no exception.”

Torrents of data

What can’t machine learning do? Promising though it is, AI has its limitations. The biggest one is managing the torrents of data that sensors deliver in ways that can be assimilated and digested to achieve practical results. As McKinsey describes it “Even more challenging in terms of scale [than developing company-wide data maintenance and governance processes], is overcoming the ‘last mile’ problem of making sure the superior insights provided by AI are instantiated [embedded] in the behaviour of the people and processes of an enterprise.”

Translated, it is hard to get everybody on board.

Despite the obvious challenges, Wartsila has seen the future and thrown itself wholeheartedly into the AI vision. In fact, it liked the product so much that it bought the company. The Helsinki-based group snapped up Transas in March 2018 as a big step forward along the journey it describes as the Smart Marine Ecosystem. In big words, Wartsila said it plans to use AI “to disrupt the industry by establishing an ecosystem that is digitally connected across the entire supply chain through applications that are secure, smart and cloud-based.”

In practical terms, Wartsila sees the vision as connecting smart vessels with smart ports in ways that “resolve inefficiencies in the shipping sector resulting from overcapacity, sub-optimal fuel consumption, and waiting times and ports and other high-traffic areas.” And that is precisely why OOCL hired Microsoft’s research tank.

It is all about converting the promise of AI into solutions in traffic control, simulators, navigation, training and fleet operation. “The combined package will further improve the way a vessel can sail in the most cost-efficient and environmentally friendly way for our customers,” summarised the president of Wartsila Marine Solutions, Roger Holm.

Few operators would disagree with that.

MSC’s virtual hostess

MSC Cruises has used artificial intelligence to create a virtual hostess whose sole job is to attend to passengers’ needs. The first cruise line to exploit AI in this way, MSC will launch the voice-activated personal assistant in 2019 on its new vessel, MSC Bellissima, currently under construction at the STX France shipyard in Saint Nazaire.

Developed by Harman International, a subsidiary of Samsung Electronics, the virtual hostess is, technically speaking, “a voice-activated conversation tool that uses AI to intelligently communicate, learn and predict the needs of passengers as well as to suggest interesting recommendations.”

That’s quite a mouthful but the best way to think of the virtual hostess is as a bit like Siri, but tailor-made for the cruise line. The as yet unnamed personal assistant is part of the cruise line’s MSC for Me, a digital innovation programme. Given the success of Siri and the imitators it has triggered, it’s certain that virtual hostesses will be appearing in competing cruise lines in the near future as they see the benefits of machine learning.

Recent whitepapers

Related articles





Knowledge bank

View all