A Study On Terrain Matching Assistant Positioning Technology For Underwater Vehicle

The dissertation makes deep researches of the key technologies of the terrain matching system of the underwater vehicle. The main work are as following:Through researching the technologies of the terrain matching system which have already existed, together with the behavior of the underwater vehicle, the terrain matching system scheme of the underwater vehicle based on ICP (Iterative Closest Point) algorithm is presented in this dissertation. This is a new way that introduces the technology of image processing to terrain matching.Firstly, it researches the mathod of gaining the digital map. The digital map is usually formed by the water depth's database measured on the seabottom. Because of the disadvantage that the database is kept secret, this dissertation introduces a new method: first, the depth data are picked up from the electronic chart, then interpolation of the depth datum by Kriging algorithm is adopted to gain the digital map. The method is the precondition of the terrain matching when it is researched by simulation, and it is the complement of the database. In this dissertation, it reviews briefly some of the popular matching primitives in computer vision and discusses their advantages and disadvantages for our matching problem. It is attractive that the contours of digital map as matching primitives are selected. Two ways of contour extraction are introduced.Whereafter, the terrain-matching algorithm based on ICP algorithm is researched comprehensively. By researching the basic theory of ICP algorithm, the key technologies of ICP algorithm are presented. The variations of ICP algorithm in the literature are summarized. Improving methods are introduced about how to solve the problem of initial value and global minimum, how to find the closest contour point, and how to find the 2D transformation of terrain matching algorithm based on ICP algorithm in the dissertation. ICP algorithm by IGG M-estimation is proposed for improving the robustness. The simulation results show that the algorithm is reasonable and effective. There are two methods about the speediness of algorithm to improve the computation. The first isselecting initial states randomly, which is possible to converge global minimum with a sufficiently small probability of error and to decrease the number of iteration. The second is computing closest point using gliding window, which can decrease the computational cost of computing closest points in the process of converging to local minima.The precision of reference navigation system is the main factor which influences the precision of terrain assistant navigation system. Especially, the reliability of reference navigation system is an important factor which influences the robustness of algorithm. The high precision and high reliability of reference navigation system is researched. First, aiming at the characteristics of miniaturization and the high precision required in the navigation system of the underwater vehicle, a new fault tolerant scheme which uses parallel filters is provided in this dissertation. The scheme not only has the optimality of centralized kalman filter but also has the characteristics of federated Kalman filter's modularity. Because the design and operation of each filter are completely independent, the fault tolerance ability is highly enhanced. Fault tolerant strategy is also introduced in this dissertation when outliers exist. This technology of fault tolerant can avoid switching the integration pattern frequently, eliminate the effection on precision by outliers and enhance the sensitivity of fault detection.In the end, the reliability and error of system, which is the most important index to evaluate the performance of navigation system, are researched. First on the foundational principle of ICP algorithm, the matching reliability of terrain assitant navigation system is researched based on intuitive geometrical reasoning, and the reliability formulas for rotation and translation are deduced. Then, based on the principle of terrain assistant navigat...