面向室内移动机器人的改进3D-NDT点云配准算法  

Improved 3D-NDT point cloud registration algorithm for indoor mobile robot

作　　者：余洪山 付强 孙健 吴司良 陈昱名 Yu Hongshan;Fu Qiang;Sun Jian;Wu Siliang;Chen Yuming(National Engineering Laboratory of Robot Visual Perception and Control Technology,College of Electrical and Information Engineering,Hunan University,Changsha 410082,China;Shenzhen Research Institute of Hunan University,Shenzhen 518057,China)

机构地区：[1]湖南大学电气与信息工程学院机器人视觉感知与控制技术国家工程实验室,长沙410082 [2]湖南大学深圳研究院,深圳518057

出　　处：《仪器仪表学报》2019年第9期151-161,共11页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金项目(61573135、U1813205);深圳市科技计划项目(JCYJ20170306141557198);湖南省科技计划项目(2018GK2021);湖南省自然科学基金项目(2017JJ3118);长沙市科技计划项目(kq1801003);航空科学基金项目(201705W1001);国家科技支撑计划项目(2015BAF11B01)资助

摘　　要：点云配准是室内移动机器人位姿估计和环境构建的关键步骤,现有点云配准算法难以工作在低纹理场景中。为提高室内移动机器人环境适应能力,提出了一种改进三维正态分布变换(3D-NDT)点云配准算法。通过改进ORB特征提取算法,确保低纹理下的特征点提取;此外,为提高点云配准精度和效率,提出改进的3D-NDT算法快速获取高精度的点云配准矩阵。采用国际知名的公共数据集TUM作为评测数据,实验结果表明本文算法达到或优于现有主流点云配准算法的性能(均方根误差低于0.02 m),相对传统3D-NDT算法配准时间缩短3倍以上;并且能工作在低纹理场景中。因此,改进的算法能提高室内移动机器人环境适应能力。Point cloud registration is a key step for indoor mobile robot pose estimation scenario building.Current point cloud registration methods hardly work in low-texture scenes.To improve scene adaptability of indoor mobile robot,this paper proposes a novel improved 3D-NDT point cloud registration algorithm this paper proposes an improved ORB algorithm to ensure valid feature extraction in lowtexture scenes;In addition,to improve accuracy and effectivity of point cloud registration,this paper proposes improved 3D-NDT algorithm for quick and accurate registration matrix solving.Quantitative results on famous TUM datasets show our system performs as good as or better than other popular solutions(lower RMSE value than 0.02 m),and time consuming decreases 3 time than traditional 3D-NDT algorithm;Notably,our algorithm can work in low texture scenes.Therefore,our algorithm can improve scene adaptability of indoor mobile robot.

关 键 词：点云配准 三维正态分布变换 三维重建 特征提取 体素网格 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]