On Seabed Terrain Aided Navigation Technology For Autonomous Underwater Vehicles

As important tools for the ocean exploitation,autonomous underwater vehicles(AUVs)have a wide range of applications in scientific research,business and military.Highly accurate autonomous underwater navigation is not only one of the key technologies for AUV development but also the hot spot and hard problem for current navigation research.Terrain aided navigation(TAN)has been considered as a potentially powerful solution for AUV navigation in GNSS-denied environments.The accuracy and viability of the TAN are affected by the availability of AUV sensors.Currently the application of traditional TAN is mature and popular on well-equipped high precision AUVs,but the TAN research is very limited on AUVs with lower sensor configuration,and in addition,high-precision navigation is an outstanding problem for such low-end system.Therefore,this work presents the TAN research on sensor-limited AUVs,including terrain modeling,terrain matching as well as TAN filter design.The major contents of this thesis are as follows:(1)To address the seabed terrain modeling,the terrain modeling based on interpolation methods is studied.A grid of bathymetric values,uniformly distributed in North and East,is chosen as a priori bathymetric map model.Different interpolation methods of the digital terrain model(DTM)are discussed,especially the modeling principle of the Gaussian Process(GP).The interpolation methods are demonstrated at two different terrain types in the priori bathymetric map.Based on the good properties of spatial correlation and uncertainty estimation in GP processing,the results demonstrate the better ability of GP to accurately represent the terrain,especially over coarse resolution map.(2)To address the large errors accumulated by DVL aided SINS(DVL/SINS)in AUVs,the seabed terrain matching navigation technology is studied.This thesis presents the constrained particle swarm optimization(CPSO)algorithm.The initial positions of particle swarm are constrained in the range of the bathymetric contour.The fitness function of CPSO is designed based on the DVL/SINS outputs and bathymetric data.The experiments compare the CPSO method and iterative closest contour point(ICCP)algorithm and demonstrate that the CPSO algorithm has advanced positioning accuracy of less than 100 m in comparison with the ICCP's positioning accuracy of more than 200 m in a rough terrain.The CPSO provides more accurate localization compared with ICCP.In addition,the study that accesses the performance of CPSO algorithm using different interpolation methods at different map resolutions further verifies that the GP-CPSO algorithm presents some advantages over sparsely or poorly mapped regions and provides better precision initial AUV position for the real-time terrain aided navigation.(3)To address the seabed terrain aided navigation for sensor-limited AUVs,the loosely-coupled TAN method and tightly-coupled TAN method are discussed in the thesis.Sensor outputs of DVL/SINS are directly put into the particle filter in the tightly-coupled TAN.The experiments compare the two coupled TAN methods and demonstrate their performance on senor-limited AUV with different sonar types(a single-beam sonar and a 4-beam DVL range sensor)in two significantly different terrain types.As a result,the performance of the tightly-coupled TAN demonstrated here is clearly a significant improvement for the navigation of sensor-limited AUV.The results further illustrate that the tightly-coupled TAN with a 4-beam DVL is able to estimate the position of the AUV with an accuracy within the horizontal resolution of the terrain map and will perform better than the tightly-coupled TAN with a single-beam sonar.(4)As discussed in the previous section,a new tightly-coupled TAN based on three-dimensional distance(3D-TAN)is presented.The method extends the one-dimensional measurement equation with only height parameter into a three-dimensional equation with horizontal distance and height.An insight into the achievable performance of the 3D-TAN filtering is demonstrated over different coarse resolution maps,different measurement errors and different terrain modeling methods.The results show that the proposed method improves TAN performance upon the performance degradation caused by the depth errors,and especially,the 3D-TAN based on GP interpolation method is more suitable for sensor-limited systems.