| 摘要: |
| 首先阐述了ORB-SLAM系统的核心思想和基本框架，介绍了SLAM系统中光束平差法的应用实现，然后针对视觉里程计部分不同的关键帧位姿优化过程分别进行了详细阐述，并利用KITTI数据集的不同序列的实验结果进行比较研究，给出了核心的光束平差法在ORB-SLAM中的具体工作方式及性能分析，最后讨论和分析了ORB-SLAM的优势和不足。 |
| 关键词: 视觉同步定位与地图构建 ORB-SLAM 光束平差法 位姿优化 |
| DOI： |
|
| 基金项目:国家重点研发计划（2016YFB0502004）；国家重点研发计划（2017YFC0821102）；北京市科技计划项目（Z171100000517006） |
|
| Overview of ORB-SLAM System Optimization Framework Analysis |
| QIU Xiao-chen,ZHAO Chen-xu,ZHANG Hai,XU Hui |
| (School of Automation Science and Electrical Engineering, Beihang University, Beijing 100083, China;Sino-French Engineer School of Beihang University, Beihang University, Beijing 100083, China;Beijing Public Transport Holdings (Group) Co., Ltd., Beijing 100161, China) |
| Abstract: |
| This paper firstly expounds the core idea, basic framework of the ORB-SLAM system and the application of bundle adjustment (BA) method in ORB-SLAM. Then, the different processes of keyframe pose optimization for the visual odometry part are elaborated separately, and the experimental results of different sequences of the KITTI dataset are analyzed and compared, the working mode and performance analysis of the core beam adjustment method in ORB-SLAM are presented. Finally, the advantages and disadvantages of ORB-SLAM are discussed and analyzed. |
| Key words: Visual Simultaneous Localization and Mapping ORB-SLAM Bundle adjustment Pose optimization |
