Safe and efficient maneuvering of a Maritime Autonomous Surface Ship (MASS) during encounters at sea: A novel approach

Abstract

When introducing a collision avoidance system on a Maritime Autonomous Surface Ship (MASS) for encounters at sea, it may seem reasonable to assume that an approaching ship will follow the procedures presented in the International Regulations for Preventing Collisions at Sea (COLREG). Experience has shown, though, that mariners sometimes, based on good seamanship, make situational adaptations where they deviate from the steering rules in the COLREG. Their goal may be to improve efficiency in own operations or simply due to courtesy of a perhaps less maneuverable approaching ship. In this paper an analysis has been performed to characterize the situations when such deviations typically occur and to estimate the conditional probability of such deviations, given the respective situation. Furthermore, the paper proposes a design for how an artificial intelligence (AI) system may handle these situations through an Extended Timeframe for AI Decision (ETAD). We suggest that the AI system search through the available clusters of situations before it normally would suggest an evasive maneuver. If the identified situation is similar to situations where the probability of an approaching ship deviating the COLREG and give way is significant, the AI could be given an extended timeframe, await its maneuver, and thereby improve the efficiency of the operation without jeopardizing the safety.