As a long time convert and user of labour saving technology, Captain Peter McArthur believes robotics and artificial intelligence have potential advantages for humanity in general and the shipping industry in particular. Where are we now and how have we got to our present situation? Where are technological advances leading our industry? What are the implications for dispute resolution?
He posed these questions at a recent London Shipping Law Centre seminar, focusing on the effect of recent and developing technology on dispute resolution. Under the chairmanship of Richard Tromans, the audience was invited to observe ’The Robot over your Shoulder.’ The seminar was hosted by HFW at their London offices.
There has been a wholesale implementation of technology in virtually every area of shipping, admittedly with outstanding benefits, particularly in cargo-related efficiency. Containers could not be managed without computers. On the downside, increasingly intelligent systems have resulted in deskilling and the increasing use of less skilled seafarers to man ships.
Capt McArthur referred to Gerd Leonard’s observation that “adoption of technology begins very slowly and then it is suddenly upon us so that, before long, we all too quickly forget how we managed without it.” This was manifest through the ongoing and tremendous increase in the number of processors in a microchip (and hence the power of the computer). Consequently, the widespread deployment of Artificial Intelligence (AI) was surely inevitable. How would this affect shipping?
McArthur referred to three technological revolutions, starting over 40 years ago.
In the first, automation systems were mainly passive feedback loops, requiring only simple programming and basic operator inputs. It followed that operators remained very much in control of outputs. There was an emphasis on labour saving and manpower productivity. Ships’ autopilots, an alternative to hand steering, were only legislated for in the 1960s. Shipboard radar had been notoriously unreliable. Even in the late 1970s, it had an unfortunate reputation for breaking down just when most needed.
In the drive towards automation and self-regulating systems in the early 1980s, engineers focused on unmanned machinery spaces. Many older generation engineers viewed this focus as the death knell for marine engineering although such fears have proved premature.
The first computers were introduced into shipping in the mid-1980s as cargo, draft survey and stability calculating tools; and for producing cargo documents, saving labour and reducing inefficiencies.
In the second revolution from the mid to late 1990s, networked computers integrated multiple data feeds through a mainframe computer which processed them according to a predetermined formula. The algorithms determined how the computer should order and process data via smaller, more reliable and more efficient processors; developments in software programming; and cheaper and near instant communications.
Global positioning systems (GPS), available only to the military barely 25 years ago, have become ubiquitous and the platform for precision navigation. When integrated with a ship’s compass, steering gear, Automated Identification System (AIS) and anti-collision radar (ARPA), GPS produces a single graphical output (ECDIS)—-supposedly everything a bridge watchkeeper might need to navigate safely and avoid collisions.
Moreover, the latest computer systems are way ahead of the most advanced anti-collision and machinery control systems currently in general use. By enlisting AI, on board systems are increasingly capable of taking autonomous anti-collision actions without involving an operator.
Video Data Recordings (VDR), fed into a simulator, recreate what the ship is doing at any given moment, which should help significantly in determining liability in an incident. Simulation activity very often forms the basis for judgements, given reasonably certainty about a ship’s position, course, speed and instructions from bridge personnel.
For expert witnesses and legal teams, multiple data feeds offer a wealth of situational and tactical information previously unavailable. Witness statements and evidence from logbooks—-possibly biased in favour of those producing them—become less relevant. If these do not tally with the VDR, they may be substantially discredited as evidence. Much of the guesswork and human intuition, therefore, has arguably been taken out of determining issues of fact. No matter how highly analysed, however, would not put an end to collisions at sea.
Capt McArthur cited further limitations of computer evidence. First, if something significant is missing from a computer program, the resultant picture of an incident may well be flawed. Further, the specialist knowledge of experienced sailors, supplemented by common sense and gut instinct, may indicate causal factors which a VDR cannot, no matter how detailed the data analysis.
Technology undermining crews and issues
Bridge watchkeeping has become a particularly sensitive area. Technology advances have led owners to argue that watchkeeping has become safer and more efficient, providing an incentive to minimise crewing levels. Onboard training is often undertaken only to meet legislative requirements. Such crews, therefore, are not trained to deal with technological problems and are less likely to know when something goes wrong, what exactly the problem is, and what to do about it.
When the Muros grounded in December 2016, the Marine Accident Investigation Branch report noted that the ship’s technology was not to be trusted. User experience was required to determine when the information displayed was reliable. The complex operator-Electronic Digital Information System (ECDIS) interface was not a unsatisfactory substitute for basic navigation skills.
Following collisions between American warships and merchant ships, the US Navy conducted a root and branch investigation into deficiencies in navigation and bridge safety procedures. They found an accident prone attitude had developed because “automated charts can lead to complacency in watch keepers.” They concluded that “for the electronic-aided system to be reliable, the operators had to understand the underlying principles of the display and question its validity when things don’t look right.”
Further, if an enemy were to disrupt a ship’s GPS system, on which navigation was completely reliant, the vessel would be effectively lost at sea and highly vulnerable. The Russians showed recently that they could interfere with GPS reception, diverting a vessel over 20 miles from its intended sea-track before anyone, relying solely on ECDIS, realised it. Accordingly, the US Navy has put fresh emphasis on teaching its navigators the basics of seamanship and celestial navigation, using the expertise of mariners trained in an earlier era.
The third revolution lies mainly in the future, with computers programmed to work with multiple variables to enable them to manage complex systems and make functional decisions. Programmers have recently developed a recursive critical network (RCN)—-whereby a computer self-learns and re-programmes itself to become more capable. Recursive capacity should enable the development of AI networks.
Three streams of automated vessels are under development:
*semi-autonomous ships with minimal crew managing the multiple automated systems onboard
*remotely operated vessels with occasional intervention by a shore-based operator
*fully autonomous vessels with no crew which would be completely self-regulating and capable of making navigational and tactical decisions without any human intervention.
The first fully autonomous vessel, the Yara Birkeland, is under construction and could start trading in 2018 along a 37-mile corridor between two or three ports in a Norwegian fjord. Initially, it will carry a commissioning crew while the concept is proved. By 2020, the Yara Birkeland should be completely unmanned and fully autonomous.
The Norwegians have passed local regulations allowing this vessel to operate in territorial waters. However, there is a huge body of regulations and international law to be addressed before such a vessel can trade in international waters.
Collision regulations will need to be completely rewritten. A massive amount of legislation and case law relating to maritime commercial practice will need revisiting. UNCLOS and individual states will need to accommodate these vessels. Pirates will probably have a field day but so will lawyers untangling hundreds of years of maritime regulation and legal precedents.
Jim Cashman, a master mariner and partner at HFW, also expressed concern about over reliance on technology in determining fact in dispute resolution.
The nature of evidence had changed over time with more information becoming available, especially through technical means. Electronic evidence was indeed a radical development. However, while all parties should be encouraged to procure and marshal such evidence about collisions and incidents, “we should not be infatuated by it.”
While the evidence of ship masters and their crew may have diminishged in relative significance, electronic devices were not necessarily well adjusted to some aspects of physical evidence on board. “Some people spend too much time on gadgets rather than looking out the window,” continued Mr Cashman. “I don’t know whether mariners will ever be replaced by robots. Either way, we are certainly a long way from it.”