海底地形匹配导航（terrain aided navigation, TAN）无需卫星或水声等精确定位手段辅助，可实现无人水下航行器（unmanned underwater vehicle, UUV）水下长期、隐蔽、精确导航。但TAN系统易在地形特征不明显区域出现匹配失效问题，通过地形特征量化分析获取地形适配性从而进行路径规划，是保障TAN系统可靠性的前提。本文针对现有地形适配性统计学变量耦合关系复杂、综合分析权重选取困难等问题，提出地形适配性的神经网络分析方法。基于历史海测子地图，通过蒙特卡罗迭代生成训练集；构建面向TAN系统地形匹配的PointNetKL网络模型，实现子地图适配性的准确、高效判断。最终，结合模板匹配算法对判断结果进行评估。本算法有效性和计算效率已通过海试数据回放试验验证。
Terrain aided navigation (TAN) can yield long-term, covert and accurate underwater navigational results for unmanned underwater vehicles (UUVs) without the supplements of accurate positioning methods, such as satellite or underwater acoustic positioning. However, TAN system might fail catastrophically in areas with poor terrain features. Thus, the calculation of TAN-suitability for areas using quantitative analysis and the subsequent path planning, is the premise to ensure the reliability of TAN systems. The complex coupling relationship of TAN-suitability statistical variables makes it difficult to select the weight of each variable in the TAN-suitability comprehensive analysis. To solve this problem, this paper proposed a TAN-suitability analysis method based on neural network. Training set are constructed using Monte-Carlo iteration with real bathymetric data. PointNetKL network was constructed to calculate TAN-suitability for areas. Finally, the PointNetKL results are evaluated using the template matching algorithm. The proposed algorithm has been proved in playback experiment using sea-trial data.
2022,44(24): 114-118 收稿日期：2022-08-18
DOI：10.3404/j.issn.1672-7649.2022.24.023
分类号：U666.11
基金项目：中央高校基本科研业务费资助项目（3072022TS0102）
作者简介：高靖萱(2002-),女,研究方向为水下无人航行器技术
参考文献：
[1] 沈鹏. AUV水下地形匹配导航的路径规划方法[D]. 哈尔滨:哈尔滨工程大学, 2016.
[2] WEI Feng, ZHE Rao, ZHONg Wang. Research on the Application of Ant Colony Algorithm in Underwater Path Planning[C]. Guangzhou, China, 2016.
[3] SHEN J, HENG L I, ZHANG J Y. Optimal Path Planning Method for Underwater Terrain-aided Navigation[J]. Torpedo Technology, 2012.
[4] LI YE, MA TENG, CHEN PENGYUN, et al. Autonomous underwater vehicle optimal path planning method for seabed terrain matching navigation[J]. Ocean engineering, 2017, 133(Mar.15):107-115
[5] MA Teng, LI Ye, ZHAO Yuxin, et al. An AUV localization and path planning algorithm for terrain-aided navigation[J]. ISA Transactions. 2020, 103:215-227.
[6] 徐晓苏, 汤郡郡, 张涛, 岳增阳. 基于熵值法赋权灰色关联决策的地形辅助导航适配区选择[J]. 中国惯性技术学报, 2015, 23(2):201-206
[7] 饶喆, 张静远, 冯炜. 一种地形匹配导航区域的可导航性评价方法[J]. 河南大学学报(自然科学版), 2016, 46(1):89-95
[8] 王立辉, 乔楠, 余乐. 水下地形导航匹配区选取的模糊推理方法[J]. 西安电子科技大学学报, 2017, 44(1):140-145
[9] SHEN J, XIONG L, XIA Q. Research on Multi-index Comprehensive Evaluation Method for Terrain Navigability of Underwater Autonomous Vehicle. 2019.
[10] BRUJIC D, RISTIC M. Monte Carlo simulation and analysis of free-form surface registration[J]. Proceedings of the Institution of Mechanical Engineers Part B Journal of Engineering Manufacture, 1997, 211(8):605-617
[11] IVERSEN T M, BUCH A G, KRAFT D. Prediction of ICP pose uncertainties using Monte Carlo simulation with synthetic depth images[C]//2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2017.
[12] BARFOOT T D, FURGALE P T. Associating Uncertainty With Three-Dimensional Poses for Use in Estimation Problems[J]. IEEE Transactions on Robotics, 2017, 30(3):679-693
[13] TORROBA I, SPRAGUE C I, BORE N, et al. PointNetKL:Deep Inference for GICP Covariance Estimation in Bathymetric SLAM[J]. IEEE Robotics and Automation Letters, 2020, PP(99):1-1
