Research On Path Planning For Autonomous Underwater Vehicle Based On Sea Map Model

Path planning is the key technique of intelligent navigation for underwater vehicle.Looking for a safe and fast navigation path in the ocean with both dynamic and static obstacles is the precondition for underwater vehicle operation. In order to plan a route of underwater vehicle in three-dimensional space environment of ocean more accurate, the study is divided into three parts. Firstly, to simulate the 3D ocean environment, divide underwater vehicle path planning problem into static global path planning and dynamic real-time obstacle avoidance. Then in the global path planning, ant colony algorithm is used to solve grid path in small grid space, and QPSO is used to plan path in large area of complex ocean environment.Finally, to estimate the dynamic obstacle motion using Kalman filtering, and to repulse away from the obstacles is based on the relative motion between underwater vehicle and obstacles.The article investigates main work as follows:By the use of water depth data in electronic chart, the underwater 3D environment model of the path planning research area is established. Extracted discrete water depth data points from 2D vector electronic chart, and original water depth data points Delaunay triangulation network model is established. A midpoint interpolation model of Delaunay triangulation is established using random midpoint displacement method based on fractal Brownian motion.Triangular inside interpolation method is put forward to lookup interpolation point fast. The method is used to change the dispersed Delaunay triangulation model to regular grid data.Improved fractal interpolation algorithm is used to change the grid data generated three-dimensional model on undersea.Ant colony algorithm is used to global path planning in submarine grid model space.After introducing the basic principle of ant colony algorithm, grid marine environment model which suitable for the ant colony algorithm to solve the path planning problem is established.According to the water depth data, the grid space could be divided into feasible space and obstacle area. Thus, the path planning problem will be converted to a set of seeking the best combination of coordinates in feasible region. The pheromones of algorithm are defined at each grid point, and the recent nine grids in the next layer are assumed to be the ant' field of view. By improving the transfer probability of ant, distance and safety of the grid and the target linear distance are considered in particle heuristic information. Computer simulations are carried out to demonstrate path planning of underwater vehicle in 3D grid submarine topography based on the ant colony algorithm, the results of the experiment are analyzed.QPSO algorithm is used in 3D path planning of underwater vehicle under complicated environment, and making a comparison between PSO and QPSO. The interpolation process is done to three-dimensional space model to obtain the higher accuracy regular grid data. The ocean current influence on the underwater vehicle navigation is discussed, and it is converted to path length which reflecting in the fitness function of particle. In order to avoid the path from being too curved, the path curvature punishment function is studied to limite the turning angle, and then the path curve transformation function is defined to reflect the path curvature in the fitness function. Design route safety testing function to deal with the generated collision path to jump out the obstacle environment. Establish a fitness function of comprehensive evaluation to evaluate the planning path. The QPSO algorithm is used to simulate underwater vehicle path planning in the complex ocean environment, and make a analysis of the experiment results.To plan a real-time dynamic path of underwater vehicle among the path points, and introduce the basic of underwater vehicle kinematics. The position data of obstacles in front of the underwater vehicle are gotting from sonar, approximating motion model of motion obstacle is established. Using the data obtained from the Kalman filter processing, in order to estimate the motion of obstacles. Establish collision prediction model and collision avoidance model which based on the relative movement between underwater vehicle and motion obstacle. The collision avoidance strategy can be divided into slowing down and turning around, and it is accomplished by controlling the line acceleration and the angle acceleration.Give each collision avoidance strategy the corresponding weights, then the solution of collision avoidance strategy model is translate into integer linear optimization problem. At last, simulation experiment is performed based on this method.