摘要
为提高地图的构建精度,提出一种混合视觉即时定位与地图构建方法。利用相机和激光雷达标定获得具有颜色信息的点云,并对点云数据进行预处理,选取相邻帧点云之间的特征点,制定匹配策略,对匹配点采取未解耦的位姿求解算法,获取两帧间的变换矩阵作为后续帧间匹配的初值。分别利用KITTI数据集和实车试验数据进行验证,结果表明,所提出的方法轨迹点均方根误差分别优化了6.84%和4.53%,验证了所提出方法的有效性。
In order to improve the mapping precision,a method of Simultaneous Localization And Mapping(SLAM)combining with camera vision is proposed.Firstly,the point cloud with color information is obtained by camera and LiDAR calibration,and then the point cloud data is pre-processed.Secondly,the feature points on neighboring frames of LiDAR scanning are selected,and then the undecoupled pose calculation algorithm is adopted for the matching points to obtain the transformation matrix between the 2 neighboring frames as the matched initial value for the next frame.Finally,the effectiveness of proposed method is verified by the KITTI dataset and actual vehicle tests respectively.The experimental results show that the performance of the root mean square error of the track points has been improved by 6.84%and 4.53%,respectively by the proposed method,effectiveness of the method is verified.
作者
张剑锋
彭育辉
郑玮鸿
周增城
林晨浩
Zhang Jianfeng;Peng Yuhui;Zheng Weihong;Zhou Zengcheng;Lin Chenhao(Fuzhou University,Fuzhou 350118)
出处
《汽车技术》
CSCD
北大核心
2021年第3期14-19,共6页
Automobile Technology
基金
福建省科技厅产学合作重大项目(2017H6007)。
关键词
激光雷达
混合视觉
即时定位与地图构建
点云
匹配
LiDAR
Hybrid vision
Simultaneous Localization And Mapping(SLAM)
Point cloud
Matching