Autonomy needs destructive testing

Malcolm Latarche
Malcolm Latarche

06 October 2016


Anxious perhaps to find a new outlet for its maritime innovation expertise, Norway has announced that an extensive area of the Trondheimsfjord in Northern Norway will be the first test bed for autonomous ships. Having pioneered the use of LNG, battery power and even to some extent fuel cells, Norway has been at the forefront of development of autonomous ships. Apparently the area was chosen because of comparatively light traffic levels and Trondheim-based Kongsberg Seatex tested various new autonomous technology solutions in Trondheimsfjord this June, together with the NTNU and the Norwegian Defense Research Establishment. The initiative was established by the Norwegian Marine Technology Research Institute (MARINTEK), the Norwegian University of Science and Technology (NTNU), the Trondheim Port Authority, KONGSBERG and Maritime Robotics. Other stakeholders include the Ocean Space Centre, and NTNU’s Center for Autonomous Operations and Services (AMOS). If autonomous ships are to feature in the future it is essential that any testing replicates the full range of situations that may be experienced during actual operations. Naturally this will be something that any flag or port state should require for obvious reasons but it is difficult to understand where the demand for autonomous ships will really come from. There are few if any systems or technologies that are entirely fool proof and electronics are perhaps more prone to failures than some would have us believe. Plane crashes, car recalls and much more have all been blamed on electronics or sensor failure at some time. There is no reason why ships should be any different and to add to the issue, autonomous ships will also need a shore-based control centre and communications input thus tripling the potential for a fault to occur. Given that possibility it must be said that the argument that autonomous ships will be safer is a somewhat dubious one. It is true that a number of incidents involving ships have been caused by human error but what is not measured is the number of potential incidents that have been avoided because of human intervention. It will be interesting to learn how many of those incidents will feature in the test programme that is to be carried out. Arguably nothing short of full destructive testing will give confidence that autonomous ships are not more of a danger than manned vessels.