参考：
Absolute Trajectory Error (ATE) 是衡量 SLAM 算法的一个重要指标。它描述了 SLAM 估计出来的机器人路径和 ground truth 之间的差异大小。显然 SLAM 算法的表现越好，那么估计出来的路径误差就会越小。
VINS-Mono is a SLAM framework developed by researchers at HKUST. The authors also published its mobile version, VINS-Mobile, which can run on iOS devices. Although they provided a pretty good README file to guide us on how to compile and run it, beginners may still feel confused. In this article, I’m going to document all the problems I met, and hopefully, it can be helpful for others.
Bundle Adjustment (BA) has a broad application in Structure From Motion (SFM) problems, which further optimize the location of points to achieve 3D reconstruction. In recent years, traditional filter-based SLAM algorithms have been taken place by optimization-based SLAM, in which BA is the core part. BA can simultaneously optimize camera poses as well as the locations of 3D points by minimizing the reprojection error.
Oriented FAST and rotated BRIEF (ORB) 特征是 SLAM 中比较常用的一种图像特征。它的准确率并没有 SIFT 高，但是其计算速度更快，可以满足实时特征提取的需求。ORB 特征还具有旋转、尺度不变性的特点，因此很适合应用在 SLAM 场景中。
Update your browser to view this website correctly. Update my browser now