On March 29th the Insurance Institute of London (IIL) held a briefing on Unmanned Ships And Their International Regulation. The speaker was Robert Veal, Research Fellow, Tutor In Law, Institute of Maritime Law at the University of Southampton.
IMN covered the meeting on March 30th as a news story. However, the topic is of much interest in the insurance sector at the moment, so we have produced a longer version. It is mainly a verbatim transcript, tidied up for grammatical purposes and for clarity in some places.
After a short introduction, Andrew Bardot, Secretary and Executive Officer of the International Group introduced Mr Veal.
Mr Bardot said: “We live in an age of acronyms and MASS, — maritime autonomous surface ships – is one we are going to have to grapple with going forward. Certainly autonomous vessels will pose both challenges and opportunities for marine insurers, and if you think that probably 70% to 80% of maritime liabilities are the result of human actions or omissions, there may be a silver lining for marine underwriters, although it may be a bit more of a headache for product liability underwriters, who may find exposures on the increase as ships become more autonomous and automated.
Technology “the easy part”:
“The technology is already well-developed, particularly in the non-commercial and military areas for quite some time, but there is a huge push towards autonomous commercial vessels, across the sectors. The technology is the easy part, I feel, at least I see it as the easy part. It’s easy to develop the technology and we see it in cars and aeroplanes and we see it in the trials of the vessels that are going to be put into commercial service.
Technology moves quickly; regulation moves slowly
I think the important thing for the insurers is to get the legal and regulatory framework that will surround the operation of these ships and what impact that will have on insurers. Technology moves quickly; regulation moves slowly. It’s good to see that the IMO has got a focus on this. They have got a paper into the Maritime Safety Committee on this subject – a review of the regulations, which is currently focused on manned ships, to see how those regulations will apply to autonomous vessels and that’s an initiative that I hope will be taken up by the IMO’s partners when they meet later in the year.
So, Robert is going to talk to us on this subject today. Robert is eminently well-qualified to do so. Since 2014 he’s been a research fellow at the Maritime Institute of the University of Southampton. He’s involved in a number of research projects and he’s made quite a number of presentations on the subject. He’s written and co-written articles. He’s a member of the CMI committee working group on unmanned ships and the law.”
Mr Veal then began his talk. Sub-headings are inserted to break up the text and to supply short “cliff notes”.
A Time of Extraordinary and Rapid Change
“Thank you very much Andrew. I think that, as Andrew touched upon, it’s fair to say that we are all living in quite an extraordinary age in which the transport sector as a whole is undergoing an exponential change.
It’s virtually unavoidable, I’m sure you will all have seen in the news, the development of driverless cars from the likes of Google, Tesla, and of course aerial drones have seemingly been around for an age.
Well the maritime domain is by no means immune to this trend. One organization seemingly at the forefront of the automated technology as far as marine is concerned is Rolls-Royce, and only recently did their vice-president reaffirm that his organization plans to introduce an unmanned passenger ferry into Finnish waters as early as 2020. And the introduction of these unmanned craft into the marine space presents unique challenges to lawyers, regulators, as well as underwriters. Today the fundamental question I shall be asking is, to what extent do these unmanned craft fall within the existing regulatory framework for shipping? And, to the extent that they do, to what extent can they comply with their rather exacting requirement?
A distinction: Remote-Controlled vs Autonomous
What the industry once called unmanned surface craft have been around for some time now. Importantly these craft are capable of self-propelled and controlled movement across the water in the absence of any personnel on board whatsoever. Now, significantly this ability is guaranteed by one of two very distinct control methods. Now on the one hand some of these craft are remote-controlled by a shoreside controller using a laptop. This is either done by using line-of-sight communication or, increasingly, the use of the global positioning system (GPS) to control craft remotely over the horizon. Imaginatively they are called remote-controlled unmanned craft. On the other hand some of these craft are instead pre-programmed and thereafter they use a combination of sonar radar, advanced computer software as well as very fast control algorithms to form a pre-determined nautical circuit without any human interaction whatsoever. These are called autonomous unmanned craft. . This binary distinction is a simplification, but for our purposes it is critical as it has such a profound bearing on the craft’s ability to meet the requirements within the existing framework. Therefore, this distinction should be borne in mind as I will be referring to it extensively as we progress.
Now today’s unmanned craft are used predominantly in the marine scientific research communities but also in the defence sector about which we know comparatively little. But they are still very small in size, very rarely larger than 10 metres in length. And it is perhaps because of this that from a regulatory standpoint they have been largely overlooked. As it stands currently, from a regulatory standpoint they are permitted, but in a limited way from a particular maritime authority. But, as I have alluded to, things are about to change, with those at the forefront of this particular technology now proposing full-sized unmanned container carriers and passenger liners. And it’s these far larger unmanned craft, exceeding 500 grt, that I will touch upon this afternoon.
Does the Existing Regulatory Framework Apply?….
“Now the real threshold question from a legal perspective is, how these craft will be regulated, and therefore “to what extent do they come within the existing maritime regulatory framework for shipping?” By ‘existing regulatory framework’ I refer to essentially two
aspects, — the international law of the sea as codified in the UN Law of The Sea Convention. I also refer to this rather lengthy corpus of IMO shipping regulations and International Civil Liability Shipping Conventions which comprise this framework. And we should also not overlook the legion of domestic national legislation which purports to transpose these international regulations into the national domain. So that is what I am referring to by the “international regulatory framework”.
… And Why It Matters
“Just before I get into the nuts and bolts about whether unmanned craft fit into this framework, let me take a second to consider why this question is so very important. In other words, what is at stake for the unmanned technology in this issue. Well, in essence it’s this. If unmanned craft do come within the existing framework, they stand to benefit in a number of vital respects.
Perhaps most importantly, from a regulatory perspective there will be no need to start from scratch, no need to reinvent the wheel in terms of their rights and obligations in the international domain. For example they already have state navigational rights and freedoms under the UN Law of The Sea Convention. Of equal importance their owners and operators will have the benefit of an international standard set of certification benchmarks in SOLAS and alike regulations. As many of you will know their owners will also have the benefit of limitation of liability under the convention of limitation of liability for maritime claims.
Conversely, if these craft do not come within this regulatory framework, they have none of these advantages. They can still attract civil liability under the ordinary legal principles, but they have no codified legal regime, no codified rights, no codified obligations. And at a stroke the timeframe for the development of a suitable legal regime will increase dramatically. So, in sum, there is awful lot at stake.
The question is, are they ships?
“So with the stakes really set out let’s address this problem head-on. Do these unmanned craft fall within the existing framework? Well, I suppose when answering this question one has to look at the existing framework and it seems to me that in all of these international maritime conventions, all the IMO regulations and also all the domestic statutes which are transposing international standards from jurisdiction to jurisdiction, for all their regional differences they essentially have one thing in common. And that’s the fact that generally speaking they all purport to apply to “ships” or to “vessels”. And therefore the question that we have to ask ourselves is: “do these unmanned craft, in legal terms at least, constitute “ships” or vessels”?
The prototypes developed by Rolls Royce (picture displayed to audience) look like ships. They perform pretty much exactly the same functions as ordinary manned ships, and because of that they will carry pretty similar operational risks to their manned counterparts. So the real question we are asking her is, is it possible to register as a ship, something which requires no seafarers to operate it? Well, when it comes to answering that question in the numerous jurisdictions I have considered it seems that it is, in principle, possible to register as a ship craft whose manoeuvrability does.
not in any way depend on seafarer operation. Of course there are numerous definitions of a ship in different legal instruments that differ subtly. But we found that, broadly speaking, this is the same across the board. And in consequence the only conclusion we can reach is that these unmanned craft do constitute unmanned “ships”, and thus fall squarely within the existing regulatory framework. And, above all, most dauntingly, have to comply with its requirements.
The salient provisions of the current framework
“So, having drawn this somewhat sobering conclusion, it really falls on us to have a look at the particularly salient provisions of the current framework to see whether these ships can comply with these requirements. In the interests of time I propose only to tackle the regulatory compliance issues in three important respects.
These are:
Crewing requirements
Collision Avoidance
Personnel training
UNCLOS
“When it comes to crewing requirements I begin with what I suppose is the foundation, the UN Convention of the Law of the Sea (UNCLOS). Indeed with Article 94, which requires that each ship be in the “charge of a Master and officers who possess appropriate qualifications in particular in seamanship, navigation, communications and marine engineering.”*
Well, for my presentation at least, what a remarkable provision that is. A requirement that every ship is in the charge of a master with appropriate qualifications. Does an unmanned ship have a master? In other words must a master necessarily be an individual discharging his offices and duties on board the ship. If we consider the remote controller of the unmanned ship at his desktop – is he the ship’s master? What about the pre-programmer of the autonomous unmanned ship? Can either of these be said to be the unmanned ship’s master? Well, needless to say tradition dictates that the master is the most senior of seafarers discharging their responsibilities on board a ship. The real question is, can the role be performed by shore-based personnel? Well this is one of very many moot points.
If you look at how the term “master” is defined in national legislation, you will find that it is often defined simply by stating that it is an individual having command or charge of a ship. Now, at least theoretically, there is no reason why this command or charge cannot be discharged remotely. If we consider the gentleman remote controlling the unmanned ship from the laptop on shore, although he is not on board he is no less in charge of the ship’s movement, its signalling, and is, I think, in a very real sense, in charge of the ship. On the other hand in an autonomous ship context, the pre-programmer is in a very different position. After setting the relevant nautical course he thereafter removes himself almost entirely from the decision-making loop, and in my view he cannot be said to be in charge of the ship. For that reason alone he cannot possibly constitute being the ship’s master. So as far as an autonomous ships are concerned, those ships in my view simply don’t have a Master, and thus fall foul of this quite essential requirement. This is really just one example of the relevant control method being deployed having a dramatic bearing on the unmanned craft’s ability to comply with international requirements. We will see that this distinction between remote control and autonomous operation continues to dictate the regulatory requirement position.
SOLAS
“Now many of those here will be familiar with the international provision for the Safety of Life at Sea (SOLAS) which prescribes a legion of technical regulations which deal with, amongst other things, ships’ navigation. Those provisions are found in chapter V, and by far the most important provision for our purposes is regulation 14, chapter V, which deals with the manning requirement of ships*. In sum it requires Contracting States to make sure their flagged ships are both sufficiently and efficiently manned. The requirement is particularized later on by a sub-section that requires this manning level to be determined by a transparent document procedure.
Now let me make one thing clear; there is no requirement in regulation 14 or in SOLAS as a whole that says at least one seafarer must be on board at any one time. There is nothing in the SOLAS regime to that effect. However regulation 14 still poses something of a conundrum for the unmanned community because, for does a requirement for manning adequacy not, of itself, implicitly prohibit unmanned ship altogether? This I think is a moot point that could probably be spun both ways.
On one side of the fence you could say that the manning adequacy assessment is always made by reference to the particular ship in question, with its individual capabilities. Therefore if you’ve got a ship that’s fitted with the most sophisticated kit, the most sophisticated communications, automated technology, which could navigate itself without any human intervention, you could argue that an onboard crew numbering zero for that ship would be technically adequate. And while that may seem far removed, you must recall that IMO guidelines on the salient points make it very plain that high levels of ship automation and shoreside support will serve to reduce a ship’s onboard manning requirements. This seems to me to be a continuation of an already-established trend.
On the other hand you could say that, as a matter of logic, if there are no people on board, that ship is not manned, and it necessarily falls foul of regulation 14 and any provision dealing with manning adequacy.
It’s a complicated question but for my own part I don’t think regulation 14 does prohibit unmanned ships. The regulation seems to me to establish a procedure or mechanism through which national administrations satisfy themselves as to the safety credentials of a ship’s proposed manning arrangements, rather than calling for any particular mode of operability. Now, having said that, I want to make one thing clear and that is not to say that any maritime administration anywhere in the world, will, or necessarily should, permit unmanned operations. On the contrary, any administration will have to be convinced as to the safety of unmanned operability, and in the absence of any specific guidelines or specific regime, tailor-made to unmanned operations, this will be an incredibly difficult task, and a long-term one. But my point is that, to the extent that they were satisfied by the safety of the technology, regulation 14 and its manning requirements would present no barrier to them.
COLREGS
“Now it follows on logically from a consideration about safe manning to one of collision avoidance and needless to say the International Regulations for Preventing Collisions at Seas 1972 (COLREGS) set out general steering and sailing directions with the aim of reducing, if not preventing outright, the number of collisions at sea. In the interests of time I propose only to deal with a very small number of these now very well-established rules. And I want to start with rule 2, titled “Responsibility”. Now, I am quite content to go out on a limb here and say that Rule 2 is the most important provision in the entire COLREGS regime. It says that ” Nothing in these Rules shall exonerate any vessel, or the owner, master or crew thereof, from the consequences of any neglect to comply with these Rules or of the neglect of any precaution which may be required by the ordinary practice of seamen, or by the special circumstances of the case.”
The principle of “good seamanship” creates problems for autonomous craft
Above all, the rule confirms the importance of good seamanship, over and above a dogmatic application of the stated letter of the rules. And, most fundamentally, it requires alternative unspecified actions to be taken where the overarching duty requires it. Now some very important consequences flow from that. What in essence Rule 2 requires is what I call “contemporaneous human sentience”. It’s a bit of a mouthful but it means real-time judgement by a person, an experienced mariner, in deciding when, on the one hand, a COLREGS prescribed manoeuvre is needed and when, on the other hand, something potentially completely different is called for.
Now this thought process poses very difficult questions for fully autonomous ships operating without human supervision. Deciding when deviation from the letter of the laws is required by this overarching seamanship duty, seems to me to be an incredibly sophisticated cognitive process, which I suggest is somewhat beyond the admittedly very impressive automated technology on the immediate horizon. In fact, I would go further and say that autonomous ships operating without any supervision simply cannot comply with this seamanship standard.
….but not for remote-controlled vessels
“Conversely, in a remote-controlled ship context, there is a human in the loop at all times, albeit onshore. It seems to me there is no reason why a seamanship standard cannot be exercised from the shore, so long as the technology enables the operator to make the required decisions and so long as the operator is appropriately trained.
This is another instance where the distinction between remote control and autonomous operation has a profound bearing on compliance.
On a related note it is traditionally a ship’s look-out which plays the integral role in informing nautical decision making, and in that regard we have rule 5 of COLREGS, which requires that “Every vessel shall at all times maintain a proper look-out by sight and hearing as well as by all available means appropriate in the prevailing circumstances and conditions so as to make a full appraisal of the situation and of the risk of collision.”
Of course traditionally watchkeeping has been performed by on-board mariners. But the real question is, can that regulation be satisfied by alternative technological means? Even today’s unmanned craft are fitted with some of the most incredibly sophisticated camera and microphone receptacles which gives the shore-based commander an incredibly sophisticated picture of the craft’s vicinity. The question is, does this satisfy the Rule 5 requirement? Well, it seems to me that the reference to “sight and hearing” again requires human perception and judgement . So in an autonomous ship context, if you have a ship that’s relying entirely on algorithmic collision avoidance, although it may have a modicum of spatial awareness, it doesn’t satisfy the rule 5 requirement of appraisal “by sight and hearing”, and it would thus fall foul of this requirement also.
The position of the remote-controlled ship is again different. As I said, the remote operator has the advantage of camera technology and can see the ship’s vicinity, albeit that the transmission is delayed to the extent that it has to come to him via satellite. But the question is whether this camera technology does satisfy the Rule 5 requirement. It’s a moot point in that the Rule does not make this clear and neither does any case law to my knowledge or any item of national legislation. But for my own part I think a fairly long history of relying on both shoreside support as well as technological advancement in satisfying this requirement indicates that any system that uses a camera to assist a shore-based operator should not be discounted or overlooked simply because it is shoreside or electronic.
As you can see from my final bullet, the rule 5 requirement is that the look-out be “proper”. So, to the extent that electronic receptacles are a substitute for a human look-out at all, their propriety I suggest would be totally dependent on things such as the quality of their recordings, their durability and of course their instantaneousness. It will be very fact-specific.
STCW
The problem of “the bridge”
“That is not the end of the matter. Watchkeeping is also dealt with in the International Convention of Standards of Training Certification and Watchkeeping, the STCW and it is chapter VIII that deals with watchkeeping and it contains a rather troublesome provision. In 8 (a) (i) there is a requirement that the officer in charge of the watch in no circumstances is to leave the bridge until properly relieved. Now, if this applies, this presents very great difficulties for both remote controlled and autonomous ships, both of which of course will probably not have an onboard bridge and in any case by definition it will not be attended by a chief officer. It could be doubted whether the STCW Convention does apply however. Article III makes it very clear that the convention shall apply to “seafarers serving on board sea-going ships”. By definition, unmanned ships have no seafarers serving on board, and thus the convention technically does not apply. Now you can say that this literal interpretation of Article III goes against the spirit of the convention somewhat, but Article III is what it is and you have to live with it. It seems to me to make the position rather plain. But for my own part I don’t think that the STCW Convention needs to apply to unmanned ships, the safety of which can be adequately addressed by the provisions of SOLAS, but to the extent that I might be wrong, and that the STCW Convention does apply to unmanned ships, paragraph 24 does need amendment before unmanned operations can go ahead.
Just lastly, I think it would be remiss not to touch on personnel training, and those present may not know that the international standards as far as manned ships are concerned are found again in the STCW Convention. Now the problem here is that we see the STCW only applies to seafarers on board, so by definition does not have an application to remote controllers or pre-programmers. But there is a contradiction here, because we have seen that a remote controller will no less have to comply with the duty of seamanship. That is made clear in UNCLOS and COLREGS. So, regardless of the STCW’s provision, those remotely controlling ships will have to undergo training at least to the extent similar to that of onboard seafarers. That of course is in addition to the training required by the new automated technology. So, as far as these new positions are concerned, for remote controllers and pre-programmers, there’s a real regulatory lacuna there as to the adequate training standards that they have to observe.
Conclusion
“Now, to offer something of a conclusion, which is incredibly difficult. I think I would sum up by saying that unmanned or autonomous technology is already in the maritime domain and it is here to stay. And that the unmanned craft that we have been talking about will almost certainly fall squarely within the existing regulatory framework for ships.
When it comes to compliance with that framework, we see that the relevant control method is absolutely critical. Autonomous operation, without any supervision by people, seems at odds with provisions not least in the COLREGS calling for the judgement of experience mariners. On the other hand remote controlled operations do not seem to be prohibited by either the letter or the principle of the current framework. I think that in either case it’s fair to say that the discretion or the roles played by maritime authorities will be critical. It will be essential that maritime authorities are convinced as to the safety of the system as a whole. And by that I mean the fidelity of the automated technology as well as the competence of the shoreside personnel.
Commercial case “by no means compelling”
“I should also say that the commercial case for unmanned ships is by no means compelling. I think that crew make up about a third of ships’ operating costs, whereas the cost of upkeep, maintaining and purchasing an unmanned ship will certainly be far higher than manned counterparts, so the jury is still very much out there.
A practical question will be will the ships be cyber-secure? How will cargo be adequately minded? How will very innocuous onboard incidents be arrested before they cause debilitating casualties. There are no real answers to that just yet. But it’s really only when this is ticked will the technology really take off.
International cooperation “essential”
“I should also say that international cooperation is going to be essential before these craft are used in world trade. That’s likely to be resisted by seafarer states who will probably resist any attempts to clarify the regime for these purposes. There’s currently very little international dialogue, let alone consensus, although some of you will know that already a group of states will be asking the IMO at its forthcoming maritime safety committee meeting to place unmanned regulation on its safety agenda. So we will look to see how that develops with great interest. But it is only when all of these complicated questions, the law, regulation, and practicality, are ticked, will the unmanned technology even start to realize is large potential.
Thank you.”
Mr Bardot thanked Mr Veal and made a quick observation, noting that “I was talking to some shipowners the other day and they said they could take out all the accommodation spaces and put on more boxes, so long-term it would be more beneficial.”
The meeting then then opened up for a Q&A session.
The first question from the floor was: “While I understand the distinction between autonomous vessels and remote-controlled vessels, do you think you may have neglected the concept of automated vessels? Because I know that the Rolls Royce initiative plans to have a fleet of automatically controlled vessels with pre-determined destinations and origins. But overlooked by a crew of sailors effectively in a desert in
Nevada operating in real time. I wondered how you thought that fits in with your thoughts on currently existing frameworks and regulations?”
Mr Veal replied: “The reality is that the binary distinction that I drew in my presentation between autonomous and remote controlled vessels was a deliberate simplification just to aid analytical expression. The reality is that the developers of the technology recognize up to 10 or 15 different levels of autonomy. It’s more of a continuum than a binary distinction. But to get back to your question. I think you are right to make the point that the distinction isn’t really between remote-controlled and autonomous operation.”
Critical Point Is Real Person ‘On Hand’ To Assume Control
Mr Veal continued: “The critical point is whether the system as a whole, be it automated or remote controlled is supervised by people who are capable of assuming remote control. The current regulatory framework already makes this requirement clear. If you look at regulation 24 of SOLAS chapter V regarding heading and track control, the ship must retain capability for a helmsperson to assume manual control immediately. So the really important distinction is between a system that’s operated by people who can assume remote control – that’s no different to being in cruise control on your car or being on autopilot on a ship, and on the other hand in other circumstances when humans are not in the loop at all and have no powers to intervene. It’s only in the latter case that I think we are having a real game-changer when we see total deference to the automated system.
Can we learn from other transport sectors’ approach to the law?
Nigel Carden, Deputy Chairman of the UK Club’s P&I management operation asked: “Can you get any help in finding approaches to changes in the law by looking at other parts of the transport industry. If we get on the DLR were there changes in the law required to apply?”
Mr Veal replied: “I think a modicum of assistance can be gleaned from looking across sector. Perhaps in particular when we are talking about general principles of the intersection between manufacturer’s liability and the owners of the relevant crafts. These principles apply in those kind of cases exactly the same across the industry. The real difficulty with maritime is that it is totally different to at least some of the other transport sectors such as road and rail because it is so international in orientation. That’s a simplification but generally it is. And the risks are very different at least from the perspective of pollution and collision and things like that. Also the maritime community recognizes rights that aren’t conspicuously available to owners in other transport sectors, not least the right of limitation of liability. So I think when it comes to broader principles I think that there is much to be gained from other sectors, but when it gets into the nuts and bolts of the regulations, unique or idiosyncratic to the marine domain, I think that very little can be gleaned at all. But I certainly watch with interest developments on the regulatory side – I think that there is a bill at the committee stage (in the UK) for driverless cars, the Vehicle Technology and Aviation Bill. It remains to be seen how far that will go. But from the point of view of broader insurer liability, perhaps it will be interesting to see where other sectors go, but there are limits to the extent that they will guide us, I think.
Technical difficulties still face autonomous operations
Mr Bardot commented: “There will obviously be many rules and regulations that will be relevant in considering the impact of automation but in the concept of negligence do you see any difficulty where this has to be detected automatically rather than by a human?
Mr Veal replied: “Well even if it were possible to put this into an automated system the problem is that you can’t sue an automated system. Generally you need people to sue for negligence.
In terms of how automation impacts on the general liability position – such as the interface between the shipowner and the manufacturer of the automated system, I do not think it is a game-changer. Even today it’s by no means easy to extract clear principles of manufacturer’s liability. E.g. how much can a shipowner be forgiven for not noticing a manufacturing flaw?
This same question may be asked in an unmanned shipping context. With, for instance, a highly complicated autonomous control system, the manufacturer probably assumes a greater level of responsibility in the circumstances, but it is by no means a green light for a shipowner not to exercise any diligence or testing.”
The next question from the floor was: “As I understand the legislation relating to driverless cars, the government is proposing that the owner is going to be held negligent. The fact that the kit broke down is his problem, not the claimant.”
Mr Veal replied: “Under the Vehicle Technology and Aviation Bill it is actually the insurer that is liable for any damage that is caused by the automated vehicle. If the vehicle is insured, then the insurer is liable.
A more detailed analysis may be found in a forthcoming paper, authored by Mr Veal, see R Veal “The integration of unmanned ships into the lex maritima” [2017] LMCLQ 303. For further information, contact [email protected]
*(It continues:,” and that the crew is appropriate in qualification and numbers for the type, size, machinery and equipment of the ship;”
(c) “that the master, officers and, to the extent appropriate, the crew are fully conversant with and required to observe the applicable international regulations concerning the safety of life at sea, the prevention of collisions, the prevention, reduction and control of marine pollution, and the maintenance of communications by radio.”