1.光照变化的SLAM数据集ICL-NUIM地址:https://cvg.ethz.ch/research/illumination-change-robust-dslam/
其数据集格式和tum相同
2.只有图片跑orb-slam2时需要:
(1)生成rgb.txt和depth.txt两个图片索引文件https://blog.csdn.net/u010128736/article/details/53079964
(2)生成rgb和depth一起的索引
python associate.py rgb.txt depth.txt生成同步的文件夹rgb_syrc、depth_syrc
associate.py文件地址:https://blog.csdn.net/Hu_weichen/article/details/91127282?spm=1001.2101.3001.6650.5&utm_medium=distribute.pc_relevant.none-task-blog-2%7Edefault%7EBlogCommendFromBaidu%7ERate-5.pc_relevant_aa&depth_1-utm_source=distribute.pc_relevant.none-task-blog-2%7Edefault%7EBlogCommendFromBaidu%7ERate-5.pc_relevant_aa&utm_relevant_index=7#commentBox
然后复制粘贴打印的信息到新建的associate.txt
(3)在终端运行,跑数据集,并可获得位姿文件。
// cd到orb_slam2目录下
./Examples/RBG-D/rgbd_tum Vocabulary/ORBvoc.txt Examples/RGB-D/tum_syn1.yaml ~/datatest/ethl_syn1/syn1 Examples/RGB-D/associations/syn1rd.txt
3.用evo测评画图 https://blog.csdn.net/qq_34213260/article/details/112548249#t4
轨迹对齐很重要:
命令 含义
–align/-a 采用SE(3) Umeyama对齐,只处理平移和旋转
–align --correct_scale/-as 采用Sim(3) Umeyama对齐,同时处理平移旋转和尺度
–correct_scale/-s 仅对齐尺度