Path Planning For Ship Navigation Based On Support Vector Machine

With the increasing of marine transportation and marine activities,it becomes an important and urgent task to improve the the automation level of ship driving.Path planning is the foundation and prerequisite for safe navigation and autonomous navigation of ships,and has important theoretical and practical value.In this paper,a path planning method based on support vector machine(SVNM)for multi-classification problems is studied.The theoretical analysis and experimental verification are carried out.Firstly,the paper summarizes the common path planning methods and the application of SVM in path planning.Secondly,this paper describes the problem of path planning,which is an optimization problem,and the method of C-SVM to solve the problem of pattern classification.Through the comparison and analysis,the feasibility and the problems to be solved by using the SVM are discussed.Thirdly,the method of solving the dual problem of convex quadratic programming problem of C-SVM is introduced in detail from the perspective of optimization theory.Through kernel trick,SVM can solve the linearly non-separable problems.In this paper,a piecewise linear SVM method with constrained Delaunay triangulation is proposed.With the combination of kernel function,the method is used to calculate each binary-class classification of SVM.Then,the method of solving the multi-classification problem by SVM is introduced,and the one versus one method is selected for path planning.The parameters of each binary-class classification of SVM are selected independently by considering the smoothness of the path and the distance to the obstacle.Finally,using C++ language in Mircrosoft Visual 2015 integrated development environment with CGAL and LIBSVM software package,the method used in this paper is verified by experiments.It includes the environment model,data preprocessing,parameter selection and calculation of paths,and gets the planning paths which are far away from obstacles and smooth.