Programming “The ordinary practice of seamen” into the AI-navigator: friendly and communicative interaction design between autonomous and manned vessels

Abstract

This paper is aimed at programmers presently being recruited to code behaviour of a new type of automatic ships capable of navigating with an unmanned bridge. Today, navigation might be summarised as “the ordinary practice of seamen”, as the collision regulations expresses it. The day after tomorrow, when all ships are automatic, sea traffic management, electronically negotiated, will ensure traffic safety and efficiency. But the challenge will be tomorrow, when automatic ships will have to coexist with traditional manned navigation. To be understandable, the mathematical algorithms governing automated ships must mimic human navigation so that a bridge officer can “read” the autonomous ship’s actions. This paper will discuss some issues concerning communicative and friendly behaviour in navigation and how mathematical interpretations of the rules of the road and seamanship will be a challenge for this new field of research. How can we design automatic behaviour that will not only be safe, but also natural and understandable for humans on remaining conventional ships, fishing boats and small leisure crafts? Artificial intelligence has the potential to handle very complex scenarios and extrapolate them further into the future than the human brain can. The risk is that this might lead to automatic manoeuvring that are counterintuitive to mariners on conventional ships. To prevent this, automation must be designed in a transparent manner focusing on clarity. And here there might be a conflict with efficiency in the sense of shortest-route and fuel economy.