Ship without a captain – what is the maritime industry working on?

In our article on autonomous systems in yachting – Autonomous systems in yachting: a step towards crewless yachts? – we showed how artificial intelligence is supporting recreational sailing today. Systems like SEA.AI help detect hazards and enhance safety on the water. However, they still play a purely supporting role, with the final decisions resting with the human operator. 

In commercial shipping, on the other hand, the scope of development is far broader. The maritime sector is working intensely on vessels equipped with systems that not only assist the crew but gradually take over operational and navigational functions. 

Built around these advancements is the concept of MASS – Maritime Autonomous Surface Ships. 

Since the subject is too vast to cover in a single article, in the forthcoming instalments of this series we will be taking a close look at the various issues relating to autonomous vessels. Today, however, we will focus on the general legal issues associated with this type of vessel. 

Are there already autonomous ships? 

A good example of the direction in which autonomous shipping is developing today is the Norwegian vessel Yara Birkeland. The vessel was designed as the world’s first fully electric container ship capable of autonomous operation. 

The main aim was to reduce road transport and shift some of the freight to sea. Every day, over a hundred diesel-powered lorries transported goods from Yara’s plants in Porsgrunn to the ports of Brevik and Larvik. Yara Birkeland was intended to gradually replace a significant proportion of these, reducing emissions of CO2 and nitrogen oxides. The project was also designed to reduce noise and dust and improve safety on local roads. 

The vessel itself is powered by an electric drive and utilises systems designed for remote and – ultimately – autonomous operation. This is made possible by the integration of sensors, monitoring and control systems, and technology that enables remote control. 

Despite the Yara Birkeland’s immense technological sophistication, some of its operations are still monitored by operators, meaning it does not yet function as a fully autonomous vessel without human supervision. Even the most advanced vessels still require a human element, and achieving full autonomy at sea is a process that will take many years. And from today’s perspective, it is difficult even to assess whether this is a realistic prospect. 

Where does the problem lie 

The greatest challenges associated with autonomous ships do not relate solely to the AI systems themselves. A far greater problem is posed by maritime law, which has always assumed – and continues to require – the presence of a human on board or within the chain of command and responsibility. 

This applies primarily to: 

• the COLREG Convention of the Prevention of Collisions at Sea,
• the UNCLOS Convention,
• the STCW Convention,
• and national maritime laws. 

The COLREG rules assume that it is a person who assesses the situation, hears the sound signals, decides on a manoeuvre and bears responsibility for the safety of navigation. 

An autonomous vessel, rather than a human, uses the following: 

• cameras,
• radar,
• sensors,
• data analysis systems,
• artificial intelligence algorithms,
• automated decision-making systems. 

Current regulations do not provide a clear answer to the question of whether an AI system can “keep watch“ or follow “good maritime practice” on its own. At this stage, therefore, it must be assumed that human involvement is necessary. 

Furthermore, at international level, there is still no standardised definition of the “master” (captain) in charge of the vessel. However, two main characteristics can be identified, namely: 

• physical presence,
• responsibility for the people on board, the cargo and the vessel itself. 

Another issue concerns obligations relating to assistance at sea. Under UNCLOS, the master of a ship is obliged to aid persons in distress. In the case of autonomous and unmanned vessels, however, the question arises: how would a ship without a crew carry out a rescue operation? 

An AI system can detect danger, send an alarm signal, notify the emergency services or change course. However, it is not capable of physically pulling a person out of the water or providing assistance in the same way as a crew on board. 

At the time the UNCLOS provisions were being drafted between 1973 and 1982, remote control of a seagoing vessel was technologically impossible. For this reason, the term “command” was sufficient to establish that the master and crew must physically be present onboard. 

The most autonomous maritime vessels already operate without the need for a human presence on board. However, as with unmanned aircraft, they are manned by people capable of taking control remotely. UNCLOS itself does not explicitly refer to physical presence of the captain and officers on board. It refers to “command”, which leaves the door open for the existence of a remote operator. 

Such supervision by a seagoing vessel’s operator would have to operate virtually round the clock to ensure constant monitoring of the navigational situation and the ability to react immediately. This model is similar to the way air traffic control towers operate in aviation. 

However, solutions of this kind raise many legal questions. These range from doubts as to which party would be liable for any damage, to the question of whether a remote operator could be regarded as the master. And – importantly – what the consequences of this would be. 

What the IMO is currently working on 

For several years now, the IMO (international Maritime Organization) has been analysing existing international conventions and assessing which provisions require amendment in light of the development of autonomous ships. 

The most important project at present is the MASS Code – a code regulating the operation of autonomous and unmanned ships. Initially, the code is intended to be non-binding, and its publication is expected in the near future. Only once practical experience has been gained will it be transformed into a mandatory international instrument. 

The work focuses primarily on: 

• maritime safety,
• rules for conducting MASS trials,
• cybersecurity,
• qualifications for remote operators,
• technical requirements for autonomous vessels,
• liability rules. 

The IMO’s plan also provides for the creation of a framework for testing and gathering experience, followed by the development of a mandatory code. Under current plans, mandatory regulations could come into force in 2032. 

Summary 

Technologies enabling partially remote and automated shipping already exist today, and the industry is working intensively on their further development. 

The coming years are unlikely to see the introduction of fully unmanned vessels operating without human supervision. A model involving a gradual increase in automation whilst maintaining control by remote operators appears to be a realistic prospect. 

The sea remains an unpredictable environment, and responsibility for the safety of shipping cannot currently be assigned to a system or an algorithm.