Research On Route Planning For Underwater Unmanned Vehicles Based On Ant Colony Algorithm

With the continuous development of modern military technology, AUV plays an increasingly important role in ocean war。Route planning is an important part of AUV’s navigation and combat. Study of Underwater Vehicle Route Planning has great significance to improve the intelligence of AUV. We can’t apply Ant System to continuous environment. In this paper, we mainly study on the route planning of AUV based on improving ant colony algorithm.Firstly, this paper introduces Ant System and several classic ant colony algorithms. Then, we study the method of building a reasonable model of ocean environment and the reasonable model of marine environment constraints. These are incorporated as constraints to route planning algorithm. Finally, the paper analyzes the problem of planning route based on Ant System in the marine environment models. To solve the problem of the huge number of points in this planning space, this paper proposes a fixed-step searching method. This method effectively reduces the size of the searching space and ensures the continuity of the searching space. We find that the visibility of Ant System fails to guide the ants. To solve this problem, we propose that the visibility is calculated based on inter-visibility analysis. This analysis takes the various constraints ocean into account. We use the Gaussian function as the model of pheromone based on the features of pheromone in nature. In order to improve the speed of ant colony algorithm and avoid the algorithm falling into local optimum, we stipulate that the smaller the cost of the route, the more the pheromone is left. We use the anti-exponential function so that the difference of each route’s pheromone increasing. To solve the problem that pheromone can’t be represented by any continuous function model, we scatter the pheromone map in order to update and store pheromone.In this paper, we designed four experiments. First we show the results in environment with a few of constrained regions. Then we analyze the performance of the visibility. The third experiments analyze the performance of the pheromone model. Finally, we show the results in environment with many constrained regions.