On ARV Integrated Navigation Technology Based On Terrain Assistance

As important underwater detection tools,Autonomous & Remotely-operated Vehicle(ARV)is one of the key underwater vehicles.The navigation and positioning accuracy of ARV is the key problem of aircraft design and development.In order to improve the positioning accuracy of ARV underwater navigation,this paper focuses on the terrain aided navigation technology.The main research and results are as follows.(1)The composition of ARV underwater terrain aided integrated navigation system is analyzed.The working principle and error model of strapdown inertial navigation system are discussed.The basic framework of underwater terrain matching problem is built and the selection of important parameters in terrain aid navigation system is analyzed.(2)To address the local modeling problem of seabed terrain,the submarine terrain modeling method based on bilinear interpolation is studied.And a seabed terrain modeling framework based on artificial fish swarm algorithm and Gaussian process regression is proposed.The good prediction ability of Gaussian process regression is utilized under the small sample and nonlinear model.Artificial fish swarm algorithm is used to obtain the optimal hyper-parameters,and finally estimates the submarine fine terrain model.(3)To address the positioning problem of ARV in the underwater straight navigation,a terrain-aided loose integrated navigation framework is built.Based on the analysis of the terrain contour matching algorithm,a loose integrated navigation algorithm based on improved terrain contour matching is proposed.The simulations show that the proposed algorithm has better navigation performance under different initial positioning errors and different flatness zone in straight-line trajectory.(4)In order to further improve the real-time performance of underwater navigation,a tightly integrated navigation framework based on terrain-aided navigation is designed.The principles of Particle filter and Gaussian sum particle filter are analyzed,and an Improved Gaussian sum particle filter is proposed to increase the positioning accuracy.The simulations show that the proposed method significantly increases on the real-time performance and precision than loose integrated navigation.Besides,the accuracy of positioning of ARV is within one resolution of the terrain map in steep area and 1.5 chart resolution in flat area.