Ship optimal path planning and artifical neural nets for berthing

In this paper, the authors present an approach based on dividing the ship control problem into two sub-problems: path planning and path tracking. A mathematical modular ship´s dynamic model is first introduced and simulations on ship maneuverability in open and shallow water are given. The problem of ship optimal path planning (SOPP) with obstacle avoidance possibilities is then formulated as a non-linear and non-convex mathematical programming problem, taking into account non-linearities, state constraints and actuator saturations. A penalty function is used to model the avoidance constraints. The discrete augmented Lagrangian approach and the Uzawa algorithm are used to solve the SOPP problem. Finally a new controller based on the artificial neural network mapping of an optimal control strategy, is used for automatic ship berthing