SNAT: Simulation-based search for navigation safety. The case of Singapore Strait

Abstract

As the bottleneck of the shipping routes from the Indian Ocean to the Pacific Ocean, the Singapore Strait is handling high daily traffic volume. In order to enhance navigational safety and reduce collisions at sea, several approaches have been proposed. However, most of the contributions adopt deterministic algorithms, failing to consider the stochasticity due to human behaviors of ship captains. Moreover, the effectiveness of these approaches is hindered by the fact that their focus is on providing a globally optimal safe set of trajectories to all vessels involved in encounter situations, almost neglecting each captain's perspective. We propose Safe Navigation Assistance Tool (SNAT), a simulation-based search algorithm to assist the captain by suggesting highly safe and robust maneuver strategies for conflict avoidance. Extensive numerical experimentation were performed proving the effectiveness of SNAT in reducing the number of conflicts, with respect to real data provided by the Automatic Identification System (AIS).