基于激光雷达的内河无人船障碍物识别方法  

Obstacle identification method based on laser radar for inland unmanned vessel

作　　者：王贵槐[1] 谢朔 柳晨光 初秀民[2] 李梓龙[2] WANG Guishu;XIE Shuo;LIU Chenguang;CHU Xiumin;LI Zilong(Wuhan Technical College of Communications,Wuhan 430065,Chian;Wuhan University of Technology,Engineering Research Center for Transportation Safe,Wuhan 430063,Chian;Wuhan University,The State Key t.aboratory of Information Engineering in Surveying,Mapping and Remote Sensing,Wuhan 430079,Chian)

机构地区：[1]武汉交通职业学院,湖北武汉430065 [2]武汉理工大学国家水运安全工程技术研究中心,湖北武汉430063 [3]武汉大学测绘遥感信息工程国家重点实验室,湖北武汉430079

出　　处：《光学技术》2018年第5期602-608,共7页Optical Technique

基　　金：湖北省教育厅重点科技项目(D20085102)

摘　　要：无人水面船（简称无人船）在内河航道环境中具有广阔的应用前景,对障碍物的智能感知是实现内河无人船自主航行的关键。针对内河无人船对近距离障碍物的感知需求,分析了内河环境下主要障碍物的激光数据特征;在对激光数据进行模式分析、数据滤波、去除聚类孤立点等预处理基础上,提出了一种基于SVM的内河典型障碍物识别方法;在所搭建的基于激光雷达的无人船环境感知系统上,进行了实验水池环境下的障碍物识别实验。实验结果表明提出的SVM障碍物感知算法识别率达到85%以上,基本满足无人船自主航行要求。The USV（unmanned surface vessels）has a wide application prospect in the inland waterway environment.The intelligent perception of obstacles is the key to realize the autonomous navigation of inland USV.Firstly,considering the short-range perception demand for inland USV,the characteristics of the laser data of the main obstacles in inland river are analyzed.Secondly,typical obstacles recognition（detection and classification）method based on SVM is put forward based on the data pre-processing,i.e.,pattern analysis,data filtering,and clustering isolation points.Finally,the obstacle recognition method is validated in the experimental pool environment,and the experimental results show that the recognition rate of the proposed method is above 85%,basically meet the requirement of unmanned navigation of inland USV.

关 键 词：无人船 障碍物识别 激光雷达 支持向量机 

分 类 号：U666.11[交通运输工程—船舶及航道工程]