一种基于拓展支持向量机的无人船路径规划方法

为有效解决无人船实现智能航行面临的路径规划问题，提出一种基于拓展支持向量机（SVM）的路径规划方法。该方法在传统SVM路径规划方法的基础上，考虑任务途中障碍物的规避风险，将工作任务的起止点作为约束条件对传统方法进行拓展，并针对不同的应用情景给出线性路径规划和非线性路径规划的求解方法。同时，为提高路径规划的效率，从SVM的原理出发，提出一种训练集裁剪的方法，大幅度减少利用SVM求解路径时所需的训练样本个数。此外，通过仿真验证该方法的可行性和有效性。

A new path planning method for autonomous ships based on extended support vector machine

As one of the most important water transportation vehicles in the future, path planning is a key technology to solve smart navigation problem of autonomous ships. To solve the path planning problem of autonomous ship, this paper develops a new path planning method based on the extended Support Vector Machine (SVM). This method is based on the conventional SVM method, which not only considers the requirement of obstacle avoidance, but also uses the start and end points of the task as constraints to extend the traditional SVM method so that the linear and nonlinear path planning solutions can be obtained for the different application scenarios. Moreover, in order to improve the efficiency of path planning, a training set clipping method is proposed to reduce the number of training samples required to solve paths for SVM. Finally, simulations are conducted to verify the feasibility and effectiveness of the new path planning methods based on the extended SVM.