Artificial assistance at the helm6 May 2022
Buoyed by advancements in AI, automation technologies and navigation systems are becoming increasingly prevalent in shipping. Usage is minimal so far, but in the future, this technology could play a vital role in reducing the risk of cruise ship collisions. Jim Banks speaks to Tony Boylen, principal specialist in assurance of autonomy at Lloyd’s Register, and Dor Raviv, co-founder and CTO of Orca AI, about potential uses.
The shipping industry is not always keen to invest in new technologies until their benefits are proved beyond all doubt, so it may be surprising that passenger ships and cargo freighters are already embracing the latest advances in automation, namely AI and machine learning.
These two entities are today’s engines of automation, as they enable systems to learn from a mass of data too large for a human brain to process and utilise, and their potential for improving performance is being leveraged in every industry sector. Currently, their use in shipping is limited and experimental, but they could soon have a dramatic impact on the safety of vessels at sea.
There are parallels with self-driving cars, which can navigate busy roads while passengers take a back seat, though fully autonomous guidance of cargo ships or cruise vessels may not be the goal for the industry just yet.
“Autonomy really relates to the navigation of a ship using digital means, a vessel that is able to undertake a predefined voyage with a reduced level of direct on-board control,” says Tony Boylen, principal specialist in assurance of autonomy at Lloyd’s Register. “There are different levels of autonomy, ranging from on-board supervision and remote watchkeeping, to remote oversight of fully autonomous functions.”
“But to achieve these different levels of autonomy we must think of the ship as an integrated system of systems,” Boylen adds. “It is not just about black-box navigation, but also about seamanship, vessel management, cargo management and the management of interactions with other vessels. And the key difference between an autonomous ship and a conventional vessel is that an autonomous ship must be able to act appropriately to protect itself, other marine users and the maritime environment in the absence of human intervention.”
“As [each of the parts of the supply chain] use more data, AI will be able to analyse this data and generate insights – both on board and the fleet level – translating into a safer, more efficient and sustainable industry.”
Autonomy in action
A technology suite that includes AI will ultimately enable navigation to be both common and repeatable, removing the vagaries associated with human beings, which include fatigue, lack of skills and poor workload management. Nevertheless, autonomy introduces other challenges.
“An array of sensor types – radar, cameras, LIDAR, AIS and others – combined with software including AI all have ambiguities that may contribute to uncertainty,” observes Boylen. “Akin to the automotive industry, the regulatory and assurance space become the critical enablers. The challenge is delivering a model that can evidentially demonstrate the safety of these solutions in a quantifiable fashion. We should recognise that autonomous cars are still not yet available for general use.
“However, there is still a great deal of confusion around autonomy, and there are incorrect visions of robot ships plying the seas without human control,” he adds. “The opposite is true. People are at the core of autonomous shipping. We rely on our seafarers’ skills to bring their ships safely into port and this will be true whether they are actually on the ships or not.”
One technology provider, Orca AI, has started to introduce AI systems for maritime vessels – notably cargo ships, ferries and tankers – to improve navigation and avoid accidents by reducing human error. Its co-founder and CTO, Dor Raviv, believes the technology will transform a very traditional industry.
“We are doing this by providing crews with unique insight, such as alerting them to dangerous targets, prioritising risk in real time and assisting with complex navigation situations,” he remarks. “This way, we enable crews to learn how their ships behave in terms of safety parameters, as well as giving them an understanding of specific areas around the world in order for them to make informed decisions when they navigate them.”
The key factor is the analysis of data to make the job of the watchkeeper easier.
“AI and automation can reduce the workload of the watchkeeper, whose job is to continuously monitor and alert on risks around the vessel, which is tedious, repetitive work,” Raviv explains. “AI, on the other hand, doesn’t get tired and never misses important information.
“The hardest use case for watchkeeping is when there are many targets surrounding the vessel,” he adds. “It’s hard to understand who is performing [the] action that could affect the safety of the ship, and AI excels in these congested areas with its ability to process multiple data points simultaneously.”
The company’s Orca AI O1 solution automatically detects, prioritises and alerts on-board crews to maritime targets in real time. It is used by some of the biggest actors in the industry, among them MSC, Maran Tankers and TMS Gas. The system integrates with all existing navigation systems on board and adds context to them via computer visualisation algorithms.
The system also continuously gathers risk data from the entire fleet and uses machine learning to identify risky patterns and present these insights to fleet managers. Raviv believes that, as such technology proves its worth, a technological revolution will happen in the shipping sector.
“The entire supply chain is undergoing one of the biggest technology transitions: the data revolution,” he says. “As [each of the parts of the supply chain] use more data, AI will be able to analyse this data and generate insights – both on board and [at] the fleet level – translating into a safer, more efficient and sustainable industry.”
“In the upcoming years, we will see more adoption of AI technologies in our everyday lives and shipping is no different,” he adds. “Since 75% of incidents at sea occur due to human error, AI will play a major role in helping seafarers gain processed information and reduce the workload.”
Crews use Orca AI to improve their communication with shore management teams, and the system has become part of daily operations for many clients. It effectively enhances their perceptibility in low-visibility conditions and assists them in analysing complex situations. In fact, clients with ships using Orca AI’s O1 system have reported a decrease in near misses compared with other vessels of the same type and operation areas.
These successes have encouraged the industry to adopt data-driven solutions faster than ever before, though bringing cutting-edge, reliable technology into the harsh marine environment remains a challenge. However, with improvements in ship-to-shore communication and the adoption of data-driven tools for fleet management, AI is likely to play an increasing role in enhancing the entire supply chain.
A framework for the future
The UK is taking the lead in providing autonomous technology, system assurance and maritime services to autonomous ships entering its ports. In 2021, the first ever unmanned marine systems certificate was awarded to SEA-KIT International by Lloyds Register, which worked with SEA-KIT on the design, operation and construction of its latest unmanned surface vessel.
“To continue this conversation around effective and safe autonomy, we need to see more examples of autonomous ships being safely trialled and demonstrated,” says Boylen. “To achieve measurable improvements in safety, seafarers need to be engaged in this process.”
Positional awareness of other vessels and the ability to identify the action or even intent of third-party vessels in the vicinity has always been a major challenge, but AI and machine learning can provide decision-making capabilities quicker than a human. With humans in the loop, actions will be taken with greater margins of safety.
“Intrinsic to autonomy is the potential ability to share highly accurate and predictive information for collaborative operation, which can collectively improve both the safety and efficiency of shipping,” continues Boylen. “This would necessitate additional layers of cybersecurity and consideration of new risks of sharing information with third parties, but if these can be overcome, similar environments would only further optimise operation.
“For [further application of AI in] larger deep-sea vessels – like cruise ships – to become a reality, the industry needs to understand the feasible use cases and commercial benefits,” he adds. “There are many examples where autonomous technology has already improved operational efficiency and safety, for example, route optimisation or collision detection systems. But the industry is yet to fully realise the benefits because they are constrained by the lack of regulatory assurance frameworks offering confidence in the technology they are being shown.”
Crucial for the cruise industry, which is making a big push for sustainability, will be the impact of AI on the use of novel fuel technologies and its ability to significantly reduce power consumption. Boylen sees this as a key capability of AI and believes that, by 2030, zero-emission autonomous ships will replace trucks when moving cargo directly from hub ports to distribution warehouses that are nearer to consumers.
“Autonomy is about starting small,” he says. “Small vessels are making big advances in survey, surveillance and maritime data capture, which is great, as small vessels consume less energy and are able to replace larger vessels for important work. This is a period of change that will be reflected by regulation, assurance and transition.
“We must recognise that a significant majority of the maritime will not make the autonomy step for some foreseeable time,” he adds. “What will be traceable is the evidence of these advanced solutions compared with the wider baseline. Simple scenarios such as a crossing event can be extrapolated into many millions of iterations. Like many technologies, confidence and familiarity will lead to wider adoption.”