Human Factors and MASS: the case for Automation Transparency

Autonomous, remote controlled, manned or unmanned ships, in IMO terminology called MASS (Maritime Autonomous Surface Ship) is on the research agenda of several countries. In Norway a 120 TEU autonomous container feeder is currently being built. Hopes are attached to safety as well as costs and efficiency benefits. The explicit assumption is that with no humans on the bridge “human error” will go away. Given that the artificial intelligence of the MASS will become fault free and follow the collisions regulations claims have been made that there should be no need to tag these ships with any signals differentiating them normal, manned ships. However, even with no navigators on the bridge of the MASS there will be many humans in the unmanned systems, so the probability is that there will be plenty of opportunities for “human errors”. There will also be the interaction between the humans on the bridge of other ships, both SOLAS and non-SOLAS vessels, and the MASS; is it possible for a properly programmed MASS to “always follow” COLREGS? And can there be “no difference in behaviour” between a MASS and a traditionally manned ship when it comes to navigation and collision avoidance? This paper will discuss some Human Factors issues that might prove challenging to the development of automatic ships.