Efficient ship weather routing using probabilistic roadmaps

Abstract

This paper introduces the probabilistic roadmap (PRM) concept in maritime transportation for solving the ship weather routing (SWR) problem. SWR is a path finding problem that seeks for the optimum route (path) for a ship moving from an initial port to a final desired port considering the weather conditions (winds, waves, sea currents). PRM is an old effective algorithm for robot motion planning based on a systematic use of a graph-based map which represents the robot's workspace. Using this map PRM explores feasible paths around the obstacles until finding a collision-free path between the starting and goal configurations of the robot. Considering the digital map of the ship's navigation area as well as forecasting data (updated every 3 h) concerning the weather conditions, a modified PRM algorithm is presented for tackling SWR. The proposed algorithm starts by taking random samples across the ship's navigation area which follow a particular distribution. Then, it uses a local planner to construct a large undirected graph (a roadmap) with nodes all the generated sample points connected with feasible (collision-free) edges (paths). Each path is related with a weight estimated by considering the weather conditions in the corresponding sea's region. Finally, given the departure and desired arrival ports, a modified Dijkstra's algorithm is applied to the resulting graph to search for the optimal path in terms of both fuel consumption and travel time that connects the two ports in the roadmap. The proposed algorithm operates in real-time determining the best solution within less than 1 min. Simulation simulations over the Aegean archipelagos as well as the Mediterranean Sea demonstrate the efficiency of the developed algorithm.