摘要
AUV要实现在未知环境中的自主导航,SLAM理论与方法是最有潜力的途径之一。本文采用前视声纳扫描水下环境,采用数字图像处理的方法对环境的声纳图像进行特征提取,并使用最邻近数据关联算法(NNF)实现数据关联,应用EKFSLAM进行水下机器人的姿态估计与环境中障碍物特征的地图构建。仿真结果表明,主动成像声纳可以成功地应用于AUV在未知环境中的自主导航;另外,通过对比点特征和线特征对AUV运行轨迹的影响,我们发现参照分布合理的点特征AUV能够更加精确地进行同时定位。
In order to achieve capabiliity of autonomous navigation for an AUV in an unknown environment, SLAM is one of the most promising ways. This paper using forwar-looking sonar to scan underwater environment, using the digital image processing to extract features from sonar images, and the rearest neighbor filter algorithm （NNF） to associate data. The EKF SLAM is applied to estimate underwater robot pose and build environment feature maps. The simulation results show that ,the active imaging sonar can be successfully applied to AUV navigation in unknown environment; In addition, by comparing the AUV trajectories generated in the cases of using lines features and points features, we found that the well-proportionedly distributed point features surrounding the AUV can be helpful of producing more precisely positioning.
出处
《微计算机信息》
2009年第4期255-257,共3页
Control & Automation
基金
国家科技部:国家高技术研究发展计划(863计划)(2006AA09Z231)"基于同时定位与地图构建方法的AUV自主导航技术"
教育部:教育部留学归国人员科研基金
"水下探测器导航与控制的关键技术(教留金2005-383)
