Path Plan of Unmanned Underwater Vehicle Using Particle Swarm Optimization

Abstract

This article proposes a method to solve route plan problem of unmanned underwater vehicle (UUV) using particle swarm optimization (PSO). Firstly, traditional electronic nautical map is preprocessed to decrease calculated amount in computation of route cost and improve plan instantaneity using technique of threaten circles covering. Secondly, choose appropriate route expression and cost computation. Lastly, route with minimum cost is received using particle swarm optimization. Result indicates that particle swarm optimization based on swarm intelligence can insure optimality most out of limited computation time, relative to traditional method using graph theory and dynamic programming. Introduction Path plan is the important content of task assignment of UUV, and purpose of path plan is to find the route from start point to goal point which satisfies UUV performance request. Generally, in the process of path plan, first step is to create a data structure represent path in the three dimension space, second step is to create appropriate cost function involve fuel, threat and etc, third step is to find optimum path using optimization algorithm which cost function achieves minimum. Swarm Intelligence is popular method to solve path plan problem in the recent years, in which particle swarm algorithm is heuristic optimization method using heuristic random search. The algorithm is started from the swarm, every particle in the swarm represents solution to UUV path plan problem. In the search process, global optimum particle sharing global optimum information with every particle and local historical optimum particle sharing local optimum information with self particle. New position of particle is addition of global optimum position and local optimum position in some extent and after some iteration particle swarm finish search to global search space. Differences of PSO and other swarm intelligence algorithm are: every particle has memory of local historical optimum solution and whole swarm has memory of global optimum solution. The principle of PSO is simple and parameters of PSO are less than other Swarm Intelligence algorithm to programming. So PSO is adapt to solve path plan problem has short time demand. This article solves path plan problem of UUV, path plan method is constituted with four steps: (1) route representation; (2) cost function choose; (3) optimum path solution using PSO. Route representation Route representation is more important problem in the path plan, which affects cost value computation and effect of path plan. Better route representation not only represents motion character, but eliminates route not satisfies plan request from standby route, and increase efficiency of PSO algorithm. Traditional method of route representation is to plot grid map from the start to the goal, which is shown in the Picture (1). Using this method, Path plan problem can be seen as shortest route on the direct map with some weight. Limitation of this method is no matter search algorithm is powerful, method can not search path point between grid points and affect optimality of UUV plan route. PSO algorithm is pointed to continue search space, and fit search map without grids, so take