High-Level Path Planning of Unmanned Sailboat for Sailing Championship and Innovative Education

In order to realize the path planning of unmanned sailboats under the influence of wind field, this paper proposes a Q-learning algorithm based on Gaussian process. Specifically, the exploratory behavior strategy of the unmanned sailboat is used to generate training samples to evaluate the value function of the optimal strategy. Secondly, the optimal value function is estimated by the Gaussian process, and the approximation of the linear function is realized. Finally, the algorithm is tested in racing scenarios and sailing around buoy scenarios. The results show that the GP-QL algorithm can well realize the path planning of unmanned sailing boats, and can be used for competitive training and innovative education of sailing boats.