利用kalibr标定IMU和相机：/ethz-asl/kalibr/wiki/Camera-IMU-calibration#2-collect-images
在开始需要明确几点的是：需要先安装kalibr，该工具主要是标定相机和IMU， 在标定的过程中需要用到IMU和相机的内参文件，这里的内参文件可以用ROS 帮忙获得，或者如果已经有这个文件可以直接用。
接下要做的即安装realsense的ros接口主要是为了方便读取topic来录制rosbag， 因为kalibr在标定的时候需要用rosbag中的数据进行标定；
这里的相机采用realsense D435i 因此在安装标定前要确保相关库已经安装
-----IMU内参标定-----
1. 确保realsense的SDK已经安装:
2. 安装realsense的ros warpper
/IntelRealSense/realsense-ros 这是前两步的参考链接，包含安装realsense的SDK和roswarpper
我之前已经安装过ros kinect在Ubuntu16.04的平台上，因此这里直接从catkin的工作空间开始
3. 标定IMU
3.1 安装ceres： /jz1570551401/article/details/78207208
1. 预安装，其中libgoogle-glog-dev 可能需要再单独安装sudo apt-get install liblapack-dev libsuitesparse-dev libcxsparse3.1.2 libgflags-dev libgoogle-glog-dev libgtest-dev 2. 下载ceres：/ceres-solver/ceres-solver3. 解压文件后进入ceres-master:mkdir buildcd buildcmake .. //如果这一步骤报错，可以检查一下是具体那个库没有安装成功，再sudo apt-get 对应的库就好make //大概等几分钟sudo make install 安装完成后/usr/local/include/ceres 和 /usr/local/lib/ 中
3.2 下载code_utils：/gaowenliang/code_utils
解压至catkin下的src文件，然后在code_utils下面找到sumpixel_test.cpp，修改#include "backward.hpp"为 #include “code_utils/backward.hpp”，然后运行catkin_make
错误提示：code_utils/backward.hpp:216:30: fatal error: elfutils/libdw.h: No such file or directory
compilation terminated.
运行：apt-getinstalllibdw-dev 安装对应的库即可，一般安装这个就可以，如果这样还不可以试试：sudo apt-get install elfutils
3.3 下载imu_utils: /gaowenliang/imu_utils
再次catkin_make
3.4 接下来就可以就可以标定IMU了：
<launch><node pkg="imu_utils" type="imu_an" name="imu_an" output="screen"><param name="imu_topic" type="string" value= "/camera/imu"/> #imu topic的名字<param name="imu_name" type="string" value= "d435i"/><param name="data_save_path" type="string" value= "$(find imu_utils)/data/d435i"/><param name="max_time_min" type="int" value= "30"/> #标定的时长<param name="max_cluster" type="int" value= "100"/></node></launch>
将该launch文件保存成imu.launch至src和install下的launch中， 然后在两个终端中同时运行
~/catkin_ws_rs$ roslaunch realsense2_camera rs_camera.launch~/catkin_ws_rs/install/share/realsense2_camera/launch$ roslaunch imu.launch
运行imu.launch文件的终端中会出现
process[imu_an-1]: started with pid [8274][ INFO] [1571305008.871258796]: Loaded imu_topic: /camera/imu[ INFO] [1571305008.872137757]: Loaded imu_name: d435i[ INFO] [1571305008.872762617]: Loaded data_save_path: /home/shinan/catkin_ws_rs/src/imu_utils/data/d435i[ INFO] [1571305008.873749085]: Loaded max_time_min: 30[ INFO] [1571305008.874493932]: Loaded max_cluster: 100gyr x num of Cluster 100gyr y num of Cluster 100gyr z num of Cluster 100acc x num of Cluster 100acc y num of Cluster 100acc z num of Cluster 100wait for imu data.
接下来等待30分钟即可，标定完成后imu的参数将会被保存在data_save_path的路径中；
也可以录制bag保存imu数据后进行标定；/fang794735225/article/details/92804030
将IMU的标定结果保存到文件imu.yaml中：
#Accelerometersaccelerometer_noise_density: 2.4753056458491202e-02 #Noise density (continuous-time)accelerometer_random_walk: 5.1092097834339175e-04 #Bias random walk#Gyroscopesgyroscope_noise_density:4.0642638745600332e-03 #Noise density (continuous-time)gyroscope_random_walk: 2.6537322064011450e-05 #Bias random walkrostopic:/imu0#the IMU ROS topicupdate_rate: 200.0#Hz (for discretization of the values above)
相机标定：
/heyijia0327/article/details/43538695
标定realsense的RGB相机，如果使用的是其他相机可以用opencv采集图像后进行单目标定，这里不多写，主要是标定Realsense的彩色相机：
设定好相机的相关参数后(主要是size 和fps)：目前设置图像大小为640*480 帧率为30
roslaunch realsense2_camera rs_camera.launch
启动realsense相机后，可以用rostopic list 查看对应的打开的topics
可以用ros自带的标定方法进行相机标定：
~/catkin_ws_rs$ rosrun camera_calibration cameracalibrator.py --size 8x6 --square 0.035 image:=/camera/color/image_raw
最开始的CALBRATE按钮为灰色，当采集的图像符合条件够就变成绿色，然后可以点击CALIBRATE进行标定，标定结果显示如下：
('D = ', [0.20848791559812113, -0.36507062055202405, 0.012079145675854016, -0.006358054188595459, 0.0])('K = ', [653.7064773464425, 0.0, 314.4404941629716, 0.0, 654.2058371422769, 261.91335661459755, 0.0, 0.0, 1.0])('R = ', [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0])('P = ', [677.4393920898438, 0.0, 311.13625111082183, 0.0, 0.0, 676.0581665039062, 266.3551872155367, 0.0, 0.0, 0.0, 1.0, 0.0])
将标定的结果保存到.yaml文件中 如下所示：
cam0: #640*480camera_model: pinholeintrinsics: [653.7064773464425, 654.2058371422769,314.4404941629716, 261.91335661459755]distortion_model: radtandistortion_coeffs: [0.20848791559812113, -0.36507062055202405, 0.012079145675854016, -0.006358054188595459]rostopic: /camera/color/image_rawresolution: [640, 480]
准备标定板文件aprilgrid.yaml：
相机+IMU标定：
/ethz-asl/kalibr/wiki/Camera-IMU-calibration#2-collect-images
1. 启动realsense相机和imu，所以也要先设置好rs_camera.launch文件中的相关参数；
其中rs_camera.launch中一定要设置如下两个参数:
<arg name="enable_sync"default="true"/>， <arg name="unite_imu_method"default="copy"/>
第一参数保证相机数据和ＩＭＵ数据的同步，第二个参数保证输出同步的ＩＭＵ数据；
2. 为相机和IMU录制数据包，
可以先运行 rosrun image_view image_view image:=/camera/color/image_raw 查看图像， 保证在录制数据包的时候标定板不会超出画面
录制的时候尽量保证图像清晰，不要剧烈移动，采集数据的起始和结束阶段注意别晃动太大，如从桌子上拿起或者放下。如果有这样的动作，在标定阶段应该跳过bag数据集的首尾的数据.同时尽可能地激活IMU的各个角度，各个方向，可以先依次绕各个轴运动，运动完后来个在空中画8字之类的操作，当然也要注意别运动太剧烈，图像都模糊了。：
rosbag record /camera/color/image_raw /camera/imu -O rgb_imu.bag
3. 进行标定：参数分别是包名称(不用加.bag)，相机参数文件，imu参数文件， 标定板文件
~/Project/kalibr_workspace$ kalibr_calibrate_imu_camera --bag ~/Project/kalibr_workspace/rgb_imu.bag--cam ~/Project/kalibr_workspace/camchain.yaml --imu ~/Project/kalibr_workspace/imu.yaml --target ~/Project/kalibr_workspace/aprilgrid.yaml
出现错误提示：
importing librariesTraceback (most recent call last):File "/home/shinan/Project/kalibr_workspace/devel/bin/kalibr_calibrate_imu_camera", line 15, in <module>exec(compile(fh.read(), python_script, 'exec'), context)File "/home/shinan/Project/kalibr_workspace/src/kalibr-master/aslam_offline_calibration/kalibr/python/kalibr_calibrate_imu_camera", line 5, in <module>from kalibr_imu_camera_calibration import *File "/home/shinan/Project/kalibr_workspace/src/kalibr-master/aslam_offline_calibration/kalibr/python/kalibr_imu_camera_calibration/__init__.py", line 4, in <module>import IccSensors as sensFile "/home/shinan/Project/kalibr_workspace/src/kalibr-master/aslam_offline_calibration/kalibr/python/kalibr_imu_camera_calibration/IccSensors.py", line 16, in <module>import scipy.optimizeImportError: No module named scipy.optimize
安装python相关库：
sudo apt-get install python-scipy
标定结果如下：
After Optimization (Results)==================Normalized Residuals----------------------------Reprojection error (cam0):mean 0.552748299525, median 0.473100512902, std: 0.3640429471Gyroscope error (imu0): mean 0.215478847361, median 0.161409013393, std: 0.19024901177Accelerometer error (imu0): mean 0.165613674102, median 0.132037510374, std: 0.129116309678Residuals----------------------------Reprojection error (cam0) [px]:mean 0.552748299525, median 0.473100512902, std: 0.3640429471Gyroscope error (imu0) [rad/s]:mean 0.0123851576362, median 0.00927736573336, std: 0.0109350130176Accelerometer error (imu0) [m/s^2]: mean 0.0579749018736, median 0.0462211936851, std: 0.04519859501Transformation T_cam0_imu0 (imu0 to cam0, T_ci): [[ 0.9999488 -0.00279248 0.00972605 0.03555196][ 0.00297818 0.99981255 -0.01913078 -0.00461578][-0.0096708 0.01915877 0.99976968 0.00838297][ 0.0.0.1. ]]cam0 to imu0 time: [s] (t_imu = t_cam + shift)-0.0280180128084IMU0:----------------------------Model: calibratedUpdate rate: 200.0Accelerometer:Noise density: 0.0247530564585 Noise density (discrete): 0.350061081538 Random walk: 0.000510920978343Gyroscope:Noise density: 0.00406426387456Noise density (discrete): 0.0574773709247 Random walk: 2.6537322064e-05T_i_b[[ 1. 0. 0. 0.][ 0. 1. 0. 0.][ 0. 0. 1. 0.][ 0. 0. 0. 1.]]time offset with respect to IMU0: 0.0 [s]Saving camera chain calibration to file: camchain-imucam-homeshinanProjectkalibr_workspacecam_imu640_480_retrgb_imu.bag.yamlSaving imu calibration to file: imu-homeshinanProjectkalibr_workspacecam_imu640_480_retrgb_imu.bag.yamlDetailed results written to file: results-imucam-homeshinanProjectkalibr_workspacecam_imu640_480_retrgb_imu.bag.txtGenerating result report...
Timeshift表示的就是IMU相对于camera的延时；
如有疑问,欢迎交流: wx:baobaohaha_ 欢迎对SLAM有兴趣的小伙伴一起交流学习~~
参考博客：/mxdsdo09/article/details/83514310
