Flight path planning of UAV based on heuristically search and genetic algorithms

Abstract

Flight path planning of UAV is a complicated optimum problem. The research in this field is usually classified as two directions: optimal path planning without considering the computation cost and real-time suboptimal path planning. Aimed at the two problems, this paper presents two methods of path planning of UAV. One method is based on heuristically search. In this method, we search the threat net by using A-star algorithm. A shortest suboptimum path is obtained, which is composed of the lines on the Voronoi diagram, and then considering the UAV's turn constraint, the path can be smoothed by geometry method. The other method is based on genetic algorithm and potential fields technology. Because potential fields, however popular, have shortcomings, viz, trap situations due to local minima, the genetic algorithm is introduced. In this method, Firstly, construct the threats' Delaunay triangle net based on the principle of nearest neighborhood and a safe planning path goes across the lines of the triangles. Then designate each point (position of a threat) on the left of the path a index 0 and on the right 1 and only designate the points of line which is not passed the same symbol. Thus, based on the directions of passing the lines of Delaunay triangle, a kind of encoding is designed on the principle of Left 0, Right 1. At last, a global planning path can obtained by the genetic algorithm and potential fields technology. At the same time, simulation results of the two path planning methods are given.