Research On Automatic Underwater Vehicle Path Planning In Unknow Environment

Automatic underwater vehicle (AUV) is a robot which works underwater automatically, which has the ability of detecting, analyzing and treating the environment information, escaping from the obstacles, tracking the targets and self navigation. The problem of the path planning is an important part in the task process, and it relates to the safe movement and navigation of the AUV. The AUV path planning is the real-time detecting local path planning according to its characters. There are a lot of researches about the real-time detecting local path planning in the robot field, but few of them can fit for the 3-D underwater complex environment. So the research of the AUV real-time detecting local path planning in the underwater environment can promote the correlative fields of the AUV. This paper focuses on the correlative parts research of the AUV real-time detecting local path planning in the underwater complex environment.Firstly, the model of 2-D sensor detection has been built after analyzing and researching the characters of the AUV detecting the environment. Meanwhile, the double safe edges approach has been proposed which can explain and analyze the real-time detected environment in the real-time detecting local path planning. This approach searches the pivotal parts of the complex environment information—double-safe-edges (DSE) information, after combining the DSE and self safe area, the DSE free work space can describe the real-time detected environment information, which can be applied to the real-time detecting local path planning.Secondly, the correlative factors of selecting the second goal points have been analyzed and researched, so the second goal points selected belief space has been proposed, which can fuse these correlative factors. This approach converts these second goal points'correlative factors into the belief functions, which can be analyzed and fused in the belief space. The analyzing and fusing process depends on the transferable belief model (TBM), which has good performance on fusing the belief and classing the targets. Thus the optimal selected second goal point in real-time detecting local path planning has been achieved. Thirdly, the joint target tracking and classification (JAC) has been analyzed and researched, and the improved method of the JAC problem base on the belief functions has been proposed. This method works on the passive bearings-only underwater target tracking system, analyzes the unscented particle filter process, and achieves the system state and measurement functions. Then the passive bearings-only underwater target tracking process base on unscented particle filter has been implemented. This process combines with the targets classification method based on TBM, so the task targets classification can be achieved, the system state function can fit for the task targets, the degree of the target classification and the precision of the target tracking have been improved, and the effectiveness of this method has been verified by simulation experiments.Finally, the double-safe-edges free work space and the correlative factors are extended to the 3-D underwater environment, so the AUV real-time detecting local path planning has been proposed. This method fits for the real-time detected and information storage in underwater environment, combining with the grid method of the sound data detected and updated. The underwater double-safe-edges free work space based on the sound data described has been built, while the movement of the AUV can be planned according to the space obstacles and the different treatment strategies of the sound data. At last, the accessibility and the security of the path planning arithmetic are analyzed as well. Some different model simulation experiments have proved the feasibility and the effectiveness of the path planning arithmetic.