预积分技术
前言
因为刚接触到SLAM这门课,里面很多的名词内部的原理不是很懂,在这里,为了避免自己写错,我先提供参考资源,后面等学习深入之后再回去补充完善笔记。
一、IMU预积分技术是什么?
RGBD-SLAM受视野范围影响,双目SLAM结构较为复杂,还是看看单目SLAM吧,但是由于单目SLAM的深度不确定性,因此还是先研究下IMU......
由于公式太多,还是直接给出网址大家去看原始文章吧.
这两篇博客写的很清楚
https://blog.csdn.net/wangshuailpp/article/details/80647621
https://blog.csdn.net/wangshuailpp/article/details/78461171
泡泡机器人上这四篇推导也很清楚
https://mp.weixin.qq.com/s?__biz=MzI5MTM1MTQwMw==&mid=2247497730&idx=1&sn=424c1527065cc9cf0a79dc160b988ab3&chksm=ec134606db64cf10a7b3e0672f4b8de6a3a4055cb21e3468d8a415134fed3d8595da2c330fd8&mpshare=1&scene=23&srcid=0727D0CPZkL2oaQDh4upwu9B#rd
https://mp.weixin.qq.com/s?__biz=MzI5MTM1MTQwMw==&mid=2247497758&idx=1&sn=3bf50d78c3a0b2490220553de4436e2c&chksm=ec13461adb64cf0ceeb8c79a922799079a940ee8762f9047829576e4c717ea0fffd24a6e9d38&mpshare=1&scene=23&srcid=0727RcCJzPyHivFkq8xlSqDI#rd
https://mp.weixin.qq.com/s?__biz=MzI5MTM1MTQwMw==&mid=2247497777&idx=1&sn=0b2bc1bbe95d3a40b4f8cd2b403fbb8d&chksm=ec134635db64cf23fcc2b8f13670a9b1c7081fa0d0c618a53aac0dcc4ab97092b2c459d40a10&mpshare=1&scene=23&srcid=0727ofEY4AqrOCbZrbEYLnbk#rd
https://mp.weixin.qq.com/s?__biz=MzI5MTM1MTQwMw==&mid=2247497852&idx=1&sn=984a7e422639e6ba818dfa97abcb5236&chksm=ec134678db64cf6e516fc67a46cd0aeecf094d36ba1eb4a209c2f5bf2efc2cea86d101b187cd&mpshare=1&scene=23&srcid=0730Jp7pvYQQgMKTdGOEgeQx#rd
一共四篇,讲的还是很不错的,推导很清楚.
还有知乎上一篇文章也不错
https://zhuanlan.zhihu.com/p/38009126
还有两篇文章也值得看看:
Forster C, Carlone L, Dellaert F, et al. On-Manifold Preintegration for Real-Time Visual–Inertial Odometry[J]. IEEE Transactions on Robotics, 2016, 33(1): 1-21.
Li J, Yang B, Huang K, et al. Robust and Efficient Visual-Inertial Odometry with Multi-plane Priors[C]//Chinese Conference on Pattern Recognition and Computer Vision (PRCV). Springer, Cham, 2019: 283-295.