Research On Underwater Vehicle Path Exploration Method In Unknown Space

In recent years,due to the important role of underwater unknown space exploration in underwater emergency search and rescue,underwater resource exploration and other engineering projects,it has gradually become a hot spot in the field of underwater research.The path search in underwater unknown space has always been a difficult problem in engineering practice.The traditional path search method is mainly based on the plane,while the underwater path search needs to be implemented in three-dimensional space.As an important carrier of underwater space exploration,underwater vehicle carrying three-dimensional scanning sonar to obtain underwater environment information is one of the most common detection modes at present.Taking the underwater unknown space vehicle path exploration as the research object,this paper studies the underwater point cloud data based vehicle path exploration method,in order to plan a reasonable and efficient dynamic path for the target search in the unknown underwater space,which has important significance and application value for underwater exploration and ocean development.The main work of this paper is as follows:(1)The preprocessing method of underwater point cloud data is studied.According to the characteristics of the underwater 3D sonar scanning mode studied,the conversion mode between local coordinate system and global coordinate system is designed,which can unify the coordinate system of the local point cloud scanned by multiple stations;Based on the characteristics of underwater point cloud data structure,octree is used to establish the topology between points to speed up the search speed;The SOR filter and ROR filter are used to denoise and compare the underwater point cloud data.After the filter processing,the underwater noise points can be effectively removed and relatively clean point cloud data can be obtained,which provides highquality data support for subsequent point cloud sequential splicing and path exploration.(2).An improved ICP algorithm based on SAC-IA is proposed.Because the sonar sensor on the underwater vehicle can only obtain the local information of the underwater environment in a single scan,it can not obtain the overall spatial information.In order to obtain an accurate and reliable environment map,it is necessary to splice the continuous point cloud images accurately and sequentially.To solve this problem,the traditional ICP registration algorithm is optimized.By establishing the FPFH feature between point clouds and introducing the pseudo Huber loss function,the robustness of the algorithm to noise points and non-coincidence points in underwater environment is improved,which can be applied to underwater map construction.(3)A path exploration method for underwater unknown space is proposed.Because the underwater vehicle has not environmental prior information in the underwater unknown space,the RRT algorithm can quickly guide the blank area in the high-dimensional environment.In view of this characteristic,a hybrid RRT algorithm is proposed based on RRT algorithm.The algorithm performs synchronous calculation according to the optimal trial of global path and local path as a whole.At the level of global path calculation,RRT algorithm is used to make coarse-grained path branch decision,and the edge signal of the selected branch is fed back to local path calculation.At the level of local path calculation,RRT* algorithm is used to search small-scale path based on peripheral detection data and frontier information;Through the rerouting operation of RRT*,it can optimize the path locally,avoid the risk of "selection shock" when it is used for overall path optimization,and finally realize the path search in underwater unknown space.The proposed method and model are verified by simulation data and measured data,which proves that the method in this paper is feasible and effective.