期刊文献+

基于改进ORB-SLAM2的单目视觉算法 被引量:1

下载PDF
导出
摘要 【目的】视觉SLAM算法受光照强弱或低纹理等场景的影响,导致产生特征提取鲁棒性差、位姿计算精度低等问题,为解决上述问题展开针对性研究。【方法】研究团队提出基于改进ORB-SLAM2的SuperPoint-SLAM算法模型,将ORBSLAM中的特征匹配模块替换成基于SuperPoint网络的特征匹配,并据此进行特征跟踪、局部建图、关键帧识别、回环检测、位姿估计。将改进后SLAM算法与ORB-SLAM2算法分别在freiburg3_sitting_xyz与freiburg3_walking_xyz_validation两个数据集进行训练验证。【结果】SuperPoint-SLAM算法在动态环境中具有比ORB-SLAM2更高的定位精度,在特征点提取数量上提升近50%,轨迹误差减少0.008 m,而且在强弱光照条件下,基于SuperPoint网络的单目SLAM相较ORB-SLAM2有更强的鲁棒性。【结论】研究团队将利用Kinect2.0,基于SuperPoint与ORB-SLAM2,进一步进行稠密地图构建,实现机器人导航。
出处 《南方农机》 2023年第13期16-19,共4页
  • 引文网络
  • 相关文献

参考文献4

二级参考文献87

  • 1周武,赵春霞.一种改进的边缘粒子滤波SLAM方法[J].华中科技大学学报(自然科学版),2008,36(S1):181-185. 被引量:4
  • 2Barshan B, Durrant-Whyte H F. Inertial navigation systems for mobile robots [ J]. IEEE Transactions on Robotics and Automa- tion,1995,11 (3) :328 -342.
  • 3Borenstein J, Feng L. Measurement and correction of systematic odometry errors in mobile robots [ J ]. IEEE Transactions on Ro- botics and Automation, 1996,12 ( 6 ) :869 -880.
  • 4Ibraheem M. Gyroscope-enhanced dead reckoning localization sys- tem for an intelligent walker[ C ]/International Conference on In- formation ,Networking and Automation(ICINA) ,2010:67-72.
  • 5Cbo B,Moon W,Seo W, et al. A dead reckoning localization sys- tem for mobile robots using inertial sensors and wheel revolution encoding [ J ]. Journal of Mechanical Science and Technology, 2011,25 ( 11 ) :2907 -2917.
  • 6Konolige K, Marder-Eppstein E, Marthi B, et al. Navigation in hy- brid metric-topological maps [ C]//2011 IEEE International Con- ference on Robotics and Automation, Shanghai, China, 2011: 3041 -3047.
  • 7Ozkil A, Fan Z, Xiao J, et al. Practical indoor mobile robot naviga- tion using hybrid maps [ C ]//Proceedings of the 2011 IEEE Inter- national Conference on Mechatronics, Istanbul, Turkey, 2011: 475 -480.
  • 8Lee S, Lim J, Cho D. General feature extraction for mapping and lo- calization of a mobile robot using sparsely sampled sonar data[J]. Advanced Robotics ,2009,23 : 1601 -1616.
  • 9Choi J, Choi M, Chung W, et al. Topological localization with kid- nap recovery using sonar grid map matching in a home environ- ment [ J ]. Robotics and Computer-Integrated Manufacturing, 2012,28 (3) :366 -374.
  • 10Liu Y, Sun Y. Mobile robot instant indoor map building and local- ization using 2D laser scanning data [ C ]//International Confer- ence on System Science and Engineering, Dalian, China,2012: 339 -344.

共引文献100

同被引文献10

引证文献1

;
使用帮助 返回顶部