Path Planning Algorithm for Unmanned Sailboat Under Different Wind Fields Based on Dueling Double Deep Q-Learning Network

The sailing path of an unmanned sailboat is restricted by the wind direction due to its own design characteristics, and how to choose the sailing direction correctly and reasonably is a problem to be solved in its path planning. Aiming at the above problems, this paper proposes an improved algorithm based on D3QN to solve the path planning problem of unmanned sailboat facing different wind directions. Firstly, this paper design a multi-objective optimization reward function for unmanned sailboat path planning based on the sailing speed, turning angle, and target orientation of the sailboat; then, design multimodal state inputs that integrate the sailboat state vector and environmental state matrix to improve the model's environmental perception ability. The simulation results have verified that the algorithm can draw reasonable navigation routes in different wind directions, which has significant advantages compared to the traditional D3QN algorithm.