Introduction

In 2019, 45% of technologists surveyed believed we’d have general artificial intelligence (AI) by 2060. When we get there, AI will be capable of independently learning to carry out new tasks that aren’t related to what it already knows how to do. But we don’t need to wait for general AI—narrow AI is already here.

Narrow AI is an AI that’s restricted to one narrowly defined task. In the same way that you can’t use the GPS to propel a ship, a narrow AI designed to identify fires on CCTV can’t optimise your engine performance. However, within those boundaries, AI provides an incredible leap forward. 

Unlike humans, computers don’t get distracted, fall asleep, or get offended if you disagree with them. They’re excellent at the boring, repetitive tasks that most humans hate. When it comes to data analysis, computers leave humans trailing far behind.

Since 2018, there’s been an 11% increase in projects and organisations claiming to use AI in their operations. Despite that, we’re only scraping the surface of the range of potential uses for AI. Even in the notoriously conservative maritime industry, we can’t escape the fact that AI is already here. This report explores the current use cases and the companies behind them.

What is AI?

At its root, AI is a computer program that aims to learn, think, and respond like a human. A normal computer program is written by a programmer who specifies how it will respond to every conceivable scenario. AI doesn’t work like that. Instead, AI identifies patterns in data and uses those patterns to work out its own responses.

Several technologies fall under the AI umbrella. In 2021, machine learning (ML) and neural networks are the most popular. ML is a type of AI known as supervised learning. To teach a ML algorithm to differentiate between container ships and tankers, you feed it training data—thousands of labelled photos of container ships and tankers—and it learns to tell them apart. The challenge is the data.

Just labelling data is labour intensive. Worse, because nearly every piece of equipment stores its data in a different place and format, accessing the data and converting it to a usable format makes it even harder.

Because of this, machine learning is limited by the practical challenges of converting and labelling data. Unsupervised learning, like that used in neural networks, overcomes this problem. It finds patterns in unlabelled data and uses these patterns to make predictions. Unlike simple machine learning, neural networks can determine whether their predictions are accurate. More importantly, they learn to improve their predictions.

Even for neural networks, the quality of the data poses a problem—if the designer isn’t careful, the AI might learn the wrong patterns. A classic case is that of the AI intended to identify wolves in photos. Instead, because all the photos of wolves contained snow, the AI learned to identify snow. 

Even worse, if there’s any bias in the training data, it also shows in the AI model. A clear example of bias was the Amazon’s CV screening AI. The training data showed that being male was an indicator of success, so the AI rejected female candidates. Similarly, a study into the use of AI to predict juvenile re-offending in Catalonia showed a strong racial and gender bias. That bias reflects an obvious pattern in the historical probation data, regardless of whether it’s a fair and accurate reflection of the actual risk of re-offending. While humans have similar bias, the impact of a biased AI in widespread use could be much larger and harder to challenge.

Supporting Technologies

Like all technology, AI relies on other technologies to function. Digitalisation, the internet of things (IoT) and big data are the main underlying technologies that support AI. Digitalisation is the process of using digitised information to improve your systems; the IoT connects physical objects, such as sensors and cameras to a network, simplifying digital data collection and transmission; and big data provides a means to process and analyse the data.

Data is the raw material that, by processing, can turn into information. The raw data is used for training AI. For our purposes, equipment and sensors generate most of the usable data. The internet of things (IoT), big data and AI are inextricably intertwined.

As discussed above, one of AI’s strengths is finding patterns in large volumes of data. This is arguably one of the most exciting areas of AI development. From identifying pirates and slave ships, to warning you when your engine needs repairs or you have a container with mis-declared cargo, it’ll be a long time before we exhaust the possibilities of big data.

Barriers and Challenges

Data Silos

If digitalisation is a key supporting technology for AI, data silos are a key barrier. Data silos result from the lack of interoperability between systems. The differing formats of data make it difficult to collaborate or use the data without first spending resources converting it to a compatible format.

Solving data silos will take a complete rethink of the culture of competition. In the short term, organisations can try to break down the silos within their organisation. Looking to the future, alliances such as the Open Industry 4.0 Alliance, supported by Lloyds Register, aim to ensure operators and OEMs collaborate, to establish a common platform and semantics, and to ensure interoperability with no vendor lock-in.

Legal Barriers

The law is another key barrier to widespread AI. On the face of it, there’s no relationship between AI and maritime law. However, the law needs to work out how to regulate AI. Law, particularly maritime law, was written with the assumption that a human would always be in the loop. With AI, that doesn’t have to be the case, but can the law keep up?

The Comite Maritime International (CMI) and the International Maritime Organisation (IMO), along with many governments, are working on regulatory scoping for marine autonomous surface ships (MASS). But there’s more to maritime AI than MASS, and in an international industry the legal grey areas can curtail its use. The absence of international agreements on topics like liability for AI-related accidents, or the approval of AI decision-support systems by some states and not others make companies rightly wary of relying on them.

Current Uses of AI in Maritime

Optimisation

The maritime industry has come a long way since the days of oars and sail. Clipper ships’ sleek hulls replaced caravels and galleons, then windjammers, with their labour-saving brace-winches and smaller crews, replaced clippers. Steam replaced sail, propellers replaced paddle wheels, HFO and diesel replaced coal. Cranes replaced union purchase, and containers replaced boxes, bags and barrels. 

Each development replaced its predecessor because it was more efficient, safer or offered cost-savings. AI optimisation follows this trend, identifying inefficiencies and suggesting better alternatives.

Condition-Based Maintenance and Engine Optimisation

Traditionally, we arrange maintenance either proactively, based on time intervals (scheduled maintenance), or reactively when something breaks. The IoT opens the door to condition-based maintenance.

Condition-based maintenance uses embedded IoT sensors to monitor equipment. An AI monitors the sensor data and alerts the operator when it detects a pattern that suggests a developing problem. The operator can then arrange maintenance. This saves time, money and spare parts, while reducing down-time.

The system can analyse the same sensor data to optimise machinery, including engines and generators, either automatically or by advising the operator. This is a popular area for AI startups, with a growing number offering related services.

Delfos is a software-as-a-service (SAAS) platform specialising in maintenance and failure prediction. Their platform uses 50 prediction models to find the best fit and provide the information necessary for optimisation.

MacGregor’s OnWatch Scout monitors and analyses data streamed from the vessel to provide guidance on maintenance and performance.

Opslock’s predictive operations tool takes a different approach. Opslok uses a machine learning model to predict workplace hazards and track planned maintenance tasks in a centralised interface.

Voyage Optimisation

Running ships is expensive, and fuel costs account for 50-60% of ship operating costs. 2019 research suggests that voyage optimisation could reduce emissions and fuel consumption by 11 to 28%, which could increase daily fleet revenue by 7.2 to 17.8%.

Are ships getting the best performance out of their engines? Which route will be the most efficient, given the draught, weather and currents? Voyage optimisation services aim to answer these questions.

Research by NAPA and Blue Insight shows that shipowners, managers and charterers are optimistic about voyage optimisation; however, they also view data validation and cost as barriers to wider adoption.

The 2020 Lloyd’s List Intelligence Big Data Award winner, Metis analyses the data from their type-approved data acquisition service to provide several optimisation services. These include route cost prediction, voyage and scenario analysis and optimisation, and charterparty monitoring. Shipowners can use the detailed metrics to inform decision-making, saving emissions and costs.

NAPA’s AI voyage planning and optimisation solution includes modules for evaluating the operational efficiency of past voyages, then uses this data to optimise the efficiency of future voyages under similar conditions. Crew on board and personnel ashore can compare route alternatives for the same voyage, improving cooperation and understanding.

Frugal Tech’s automated engine optimisation system can be retrofitted to existing ships. It uses a wide range of sensors to optimise RPM and propeller pitch. They guarantee 10% fuel savings on most vessel types. They also have built-in speed and power pilots that maintain specified speed-over-ground or set power level to meet charterparty requirements.

Founded in Germany, Deep Blue Globe’s ECDIS-compatible system uses real-time satellite data from the EC’s Copernicus program to optimise every aspect of a ship’s journey. They claim 5-10% fuel cost savings, improved safety, shorter voyages and more accurate ETAs.

Searoutes combines historical and real-time data, weather data, and ship performance with their ML system to calculate accurate sea distances and routes. Subscribers can use this information to improve their decision-making, calculate accurate ETAs, monitor fuel consumption, estimate emissions, and compute deviation costs.

Ociana from GSTS processes satellite data with data from ocean, weather and port activity sources to provide decision-making information. This doesn’t optimise just fuel consumption and emissions, but also operations and logistics. The customisable alerts can flag illegal and irregular activities and non-compliance, supports maritime security, improve ETA management and minimise congestion in restricted waters.

Port Call and Logistics Optimisation

Regardless of whether you look at it from the perspective of the ship’s crew, the port, or the charterer, it’s in everyone’s interest to make port calls as efficient and stress-free as possible. Most stakeholders realise that this is not currently the case.

Non-standard paperwork, scheduling, cargo operations, bunkers and stores, pilotage, tugs, berth allocation, security, customs, immigration and more make port calls a challenging and inefficient process. Improvements in any area will improve efficiency and safety, lower costs and emissions, and reduce stress for all involved. 

A 2017 study found that advising ships of berth availability twelve hours before arrival at the Port of Rotterdam could save 134,000 tonnes of CO2 emissions a year. The Port Call Optimisation Task Force showed the wide range of tasks and participants involved in a port call. Fortunately, many companies are working on the problem.

With Sinay’s estimated time of arrival (ETA) module, ports and shipping managers can use historical AIS data to predict vessel routes and calculate accurate ETAs. More accurate information helps with planning and scheduling, which helps to avoid congestion, improve sustainability and reduce costs.

Trusted Docks provides a different of service, bringing AI recommendations to the maritime industry. Trusted Docks’ recommendation engine uses a combination of AI and human experience to make it easy to choose shipyards and repair facilities, see other companies’ reviews, and generate ship repair projects and docking requests.

AiDock’s AI “virtual assistants” automate all import and export paperwork, including pre-alert documents, tax, customs, critical shipments, and rate and route analysis. This helps courier companies, freight forwarders, postal services, and customs authorities boost productivity, and lets staff focus on customer service and business growth.

The Prephall AI-powered chatbot makes it easy for consumers to get answers fast. Companies can automate their quotation process, automate booking processes, notification and data collection, and easily track shipments.

The Secondmind Decision Engine provides custom AI solutions to help people make complex decisions. In the transportation and logistics sector, they offer demand forecasting and planning, asset allocation, dynamic pricing, and network and inventory optimization.

Transmetrics provides cargo companies with data cleaning, demand forecasting and predictive optimisation and execution based on data science and AI. Transmetrics’ insights have saved their customers up to 25% in linehaul costs. From their start in Bulgaria, they’ve expanded to worldwide operations, with services including asset positioning and last-mile, linehaul and maintenance planning.

Autonomous Ships

Marine autonomous surface ships are usually known by the acronym MASS. While national laws and international conventions lag behind, MASS companies are surging ahead. People have many opinions about MASS, both for and against, but whether you like it or not, it’s here. People often conflate MASS and unmanned ships. These aren’t the same thing, and MASS-related technology has a lot to offer manned ships as well.

The International Maritime Organisation (IMO) has defined four levels of autonomy:

1. Ship with automated processes and decision support: seafarers are on board to operate and control shipboard systems and functions. Some operations may be automated and occasionally unsupervised, but with seafarers on board ready to take control.
2. Remotely controlled ship with seafarers on board: the ship is controlled and operated from another location. Seafarers are available on board to take control and to operate the shipboard systems and functions.
3. Remotely controlled ship without seafarers on board: the ship is controlled and operated from another location. There are no seafarers on board.
4. Fully autonomous ship: the operating system of the ship is able to make decisions and determine actions by itself.

Level 1 has the most to offer manned ships in the short-term, with decision support systems. Like bridge watchkeeping officers, autonomous ships will need to detect targets around them and decide on a course of action. While not perfect, computers outdo humans in staying awake and focused.

The features work in much the same way on both manned and unmanned vessels. On manned vessels, the decision-support system provides the information to the bridge watchkeeper to assess and implement, while on unmanned ships the AI sends the appropriate signals directly to the vessel’s steering and propulsion systems.

Guardian AI is a powerful AI system for the safety and control of both manned and unmanned vessels. The machine learning and computer vision systems are built on scalable sensor analytics and provide accurate vessel control, even on low-bandwidth systems. The system offers floating object detection, identification and analysis, as well as advice on collision avoidance, route planning, weather avoidance and more.

Captain AI uses high-fidelity simulation, cutting-edge sensors and state-of-the-art deep learning techniques to develop a safe and fully autonomous shipping solution.

Buffalo Automation offers autonomous navigation technology for both commercial and recreational vessels. Their systems include autonomous navigation, situational awareness, decision support, remote monitoring, data analytics and cloud-based fleet management. Their small autonomous demonstration vessel can avoid obstacles while manoeuvring at up to 30 knots.

Marine autonomy isn’t restricted to surface ships. Rovco’s SubSLAM uses state-of-the-art solutions to provide high-quality hydrographic survey data using 3D and Artificial Intelligence.

Maritime Robotics offers several options, including a conversion system which can convert a manned to an unmanned vessel. Users can take advantage of the possibilities of unmanned surface operations, while still having the option of manned use.

Sea Machines’ industrial-grade systems can be installed aboard existing commercial vessels, or included in new-build packages. They’re designed to enable remote and autonomous operations of all types of commercial vessels.

Robotics

Firefighting, tank and damage surveys can be dangerous. Underwater hull surveys, oil spill clean-up and hull and tank cleaning are difficult and expensive. Extended search-and-rescue operations in tanks can cost lives, and hold inspections are tiring and time-consuming. But robots can change this.

With robots, people can stay at a safe distance from dangerous or unpleasant tasks. Robots reduce exposure to physical hazards, chemicals and dangerous atmospheres. Robotics can work without AI; however, when they work together, they form an effective combination.

Sea Robotics hullBUG is a semi-autonomous cleaning system that uses light brushes to clean a ship’s hull and remove the early signs of biofouling. A clean hull improves efficiency and reduces emissions; being able to do this in the water saves time and costs when it comes to dry-docking. Their tankBUG works in a similar way in tanks, but also identifies any hidden corrosion or structural compromise, avoiding costly maintenance or repairs.

Built on a modular core platform, Imotus Hovering Autonomous Underwater Vehicle can perform inspections in complex confined spaces, or resident missions in open water. It uses ROS over ethernet to communicate, and proprietary algorithms to enable autonomous operations in a variety of confined space environments.

While not yet available commercially, the University of the Balearic Islands developed the ROBINS autonomous drone for cargo hold inspection. It uses a range of sensors and software technology for autonomous unsupervised navigation and can explore and inspect open spaces like holds and tanks.

Cargo

After container fires led to several deaths, a National Cargo Bureau survey found 55% of containers failed inspection with one or more deficiencies, including misdeclared or improperly stowed cargo. Of import dangerous goods containers, 69% failed, along with 38% of export dangerous goods containers.

Hazcheck Detect scans all booking details for keywords, validates against rules and highlights suspicious bookings to identify misdeclared and undeclared dangerous goods and other compliance cargo. This avoids misdeclared and undeclared cargo from being loaded onto ships, reducing the risk of fire.

Tracking and Surveillance

It’s undeniable that tracking and surveillance have sinister connotations and have been misused by many organisations throughout history. However, it’s equally inescapable that they have the potential to be a huge step forward in safety and security. Both at the individual crew level and the international level, AI tracking and surveillance alerts can reduce injuries, marine casualties and security incidents. 

While regulations and procedures have come a long way in improving safety, they will never prevent every problem. Flagging personnel who fall overboard, tracking passengers in emergencies, identifying pirates and vessels in distress, stopping fires as soon as they start, and flagging stressed or fatigued staff are just a few of the current AI tracking and surveillance tools in use in the maritime industry.

Tracking Ships

In 2019, the Japanese government announced a research program to bolster AI maritime surveillance using aircraft. The program will allow them to expand their Maritime Self-Defence Force’s territorial monitoring, with potential security and safety improvements.

Whether or not you’re interested in identifying illegal fishing boats, slave ships and vessels in distress is a matter of your moral values. However, if you’re a seafarer working in a pirate area—or their insurer—identifying pirates is a matter of intense interest. The thing is, all three tools rely on the same technology and very similar data.

Global Fishing Watch uses a neural network to process a combination of satellite and AIS data. The AI knows the patterns of behaviour exhibited by boats engaged in illegal fishing; when it sees those patterns it flags the boat. But they also search for the patterns of behaviour exhibited by boats manned by slaves. 

If AI can use ships’ behaviour to identify illegal fishing and slavery, researchers realised it could detect refugees and pirates in the same way. 

ST Engineering’s Maritime Anti-Piracy System (MAPS) analyses the data from the ship’s CCTV and track-before-detect radar (TBD), and provides automatic detection and alert triggering of suspicious vessels. The technology hinges on the smart analytics of ship courses, speed and movement patterns. MAPS provides early warning on suspicious vessels through smart mobility tracking and in-depth behavioural analysis. 

DeepSea Technologies offers AI-powered vessel and fleet tracking and optimisation. At a ship level, ship-specific route, speed and trim optimisation, real-time tracking and performance reporting minimise costs and maximise efficiency; at a fleet level, automated analysis helps companies understand the performance of every vessel and how it compares with the global fleet.

Researchers are still looking into new uses for ship-tracking data. A 2018 paper presented at the IEEE International Conference on Data Science and Advanced Analytics (DSAA) uses the same data to describe and quantify search and rescue operations in the Mediterranean Sea.

When the UN Statistics Agency made a large body of AIS data available, researchers used it for an incredible variety of projects. These included calculating time in port and port traffic, tracking trade flows, and NOx, SOx and CO2 calculation. While these aren’t yet commercially available, they show the range of possibilities for ship tracking data.

Tracking People

“Big brother is watching.” Trust is the biggest challenge with any technology that tracks people. There are very few people who are comfortable discussing emotions with their employer, and tracking non-employees comes with consent issues. It’s a minefield of ethics, cyber-security, data protection and more. However, for companies that can overcome the barriers, AI surveillance technologies offer a significant safety improvement.

Dolphin Vision offers a flexible AI-based visual surveillance solution. Their combination of advanced computer vision and machine-learning techniques can detect and analyse people’s behaviour in complex environments. In the maritime industry, this could be useful for a wide range of tasks as diverse as port security and drowning detection.

Senseye uses off-the-shelf cameras to monitor brain activity by tracking eye movements. This can tell whether a user is fit for duty by checking for fatigue, intoxication, or stress impairment. In an industry where stress and fatigue can have immediate and severe consequences, this real-time assessment helps reduce the risks and keep your people safe. The same data can provide other insights including improved telemedicine, mental health diagnosis and personalised e-learning. 

Sensing Feeling’s deep-learning technology processes video in real-time to provide accurate, behaviour detection of groups of people. The system can provide specific alerts when it detects pre-determined behaviours, such as the early warning signs of stress and physical fatigue. Supervisors can monitor situational risks in real-time to reduce accidents and near-misses in safety-critical environments such as ships and ports.

Tracking Cargo and Equipment

Container yards are complex. Keeping track of the containers, ensuring they’re stacked in the right place, in the right order, moved to the right ship at the right time—or the reverse is a constant challenge. Just one mistake can cause costly delays and frustration. 

eYard’s AI optimises container position, reduces the number of avoidable—and costly—container-moves, and improves efficiency. Their smart reporting gives management a clear overview of their operations, allowing more informed decisions.

Fike Video Analytics’ video-based flame, smoke and oil mist detection system uses state-of-the-art analytics to detect the problem at its source and raises the alarm, saving crucial seconds on response.

Risk Assessment and Management

The maritime industry, like so many others, is a constant balancing act between risk and reward. Whether the risks are physical, environmental, commercial, or financial, get it wrong at your peril.

Windward helps insurers and businesses, governments and other organisations understand, manage and reduce maritime risk. Their data-driven risk insights range from commodity trading and shipping to customs, defence and intelligence. They’re based on actual ship operations and provide immediate go/no-go recommendations to help customers make the best decisions.

Concirrus’ insurance analytics provide a scalable, data-driven, real-time risk management platform that ensures customers can review and respond to changes in exposure as they occur. Their Quest products combine historical claims information with static demographic and dynamic behaviour-based data sets to reveal the behaviours that correlate to claims. The outcome is new insights and rating factors that simply did not exist before, the ability to better deploy risk capital, improve loss ratios and drive down operating costs.

Future Uses of AI in Maritime

Medical Care

Ships with over 100 persons on board at sea for three days or more carry a doctor. Other ships rely on the crew to deal with medical emergencies on board. Senior deck and engineering officers receive a whole week of medical training; junior officers get three days. With this training and a radio, they’re expected to deal with all on-board medical emergencies.

As technology and on-board communication have progressed, telemedicine has improved. First, email and satellite phones replaced radios; now AI is on the scene.

The Maritime Labour Convention (MLC) requires that, “…seafarers are given health protection and medical care as comparable as possible to that which is generally available to workers ashore, including prompt access to the necessary medicines, medical equipment and facilities for diagnosis and treatment and to medical information and expertise.”

From cancer diagnosis to natural-language medical chatbots, the use of AI in healthcare ashore is increasing. Despite attempts by several startups, it isn’t used in maritime medical care. However, given the MLC requirements and the high cost of medical diversions, it’s only a matter of time before AI-supported telemedicine comes into use at sea. 

Stability

For ships, stability is critical. On most ships, officers calculate stability before departure based on reported or estimated cargo weights. Incorrectly declared weights, miscalculated stability, and commercial pressure lead to many serious marine accidents.

While some companies are working on identifying mis-declared cargo, others are working on systems to manage the stability. Ships can measure their stability with an inclining experiment. In a nutshell, they move a weight to one side and see how much the ship tilts. 

Tymor Marine’s intelligent stability technology can carry out an automated inclining experiment to calculate the ship’s stability, saving the officers time and effort, and improving safety. In the future, similar integrated AI systems could use real-time ship movement data to manage stability with no human input.

Autonomous Rescue Vessels

ASV and the University of Portsmouth are working on an autonomous rescue vessel. Once complete, it will automatically launch itself when a person falls into the water. Both on shoreside facilities and on ships, this offers a considerable improvement over the current system.

Legal Advice for Masters

In UK law and by custom, the ship’s master is, “master under God.” The master is legally responsible for everything on board, including the condition of the ship, care of the cargo, safety of the crew, interactions with authorities, and compliance with the ever-expanding body of laws, conventions, rules, recommendations and regulations.

Ashore, maritime law and commercial law are specialist areas that require extensive training; masters often have to make legal decisions based on their best guess and live with the consequences.

Lawyers already use AI. It helps them perform due diligence and research, provide insights, and automate processes. Even simple chatbots can help laypeople with legal decisions. While maritime again lags behind industry ashore, given the complex legal environment faced by ship masters, this is an area ripe for disruption.

Conclusion

We’re living in a future that just ten years ago was the stuff of science-fiction. Both in maritime and ashore, AI is developing faster than most of us ever imagined. Because of its international nature, the maritime industry faces unique barriers to AI adoption; however, as the technology matures, its use will expand to encompass every aspect of the industry, bringing improvements in safety and ease-of-use.

When Steve Jobs unveiled the iPhone in 2007, no-one foresaw the countless uses for today’s smartphones. In the coming years, we can expect to see AI become as familiar as smartphones. Particularly at sea, the biggest barriers to AI adoption are not technical or legal: they’re human. 

There’s a common quip that seafaring is the second-oldest profession. Whether or not it’s true, the industry has been around for a long time and is mired in tradition. Over the years those traditions, the time-tested ways of doing things, have sustained the industry and saved countless lives; however, unless we address the industry’s resistance to change, those traditions will become an anchor holding us back.