利用kalibr標定IMU和相機： https://github.com/ethz-asl/kalibr/wiki/Camera-IMU-calibration#2-collect-images
在開始需要明確幾點的是：需要先安裝kalibr，該工具主要是標定相機和IMU， 在標定的過程中需要用到IMU和相機的內參文件，這裏的內參文件可以用ROS 幫忙獲得，或者如果已經有這個文件可以直接用。
接下要做的即安裝realsense的ros接口主要是爲了方便讀取topic來錄製rosbag， 因爲kalibr在標定的時候需要用rosbag中的數據進行標定；
這裏的相機採用realsense d435i 因此在安裝標定前要確保相關庫已經安裝
-----IMU內參標定-----
1. 確保realsense的SDK已經安裝:
2. 安裝realsense的ros warpper
https://github.com/IntelRealSense/realsense-ros 這是前兩步的參考鏈接，包含安裝realsense的SDK和roswarpper
我之前已經安裝過ros kinect在Ubuntu16.04的平臺上，因此這裏直接從catkin的工作空間開始
3. 標定IMU
3.1 安裝ceres： https://blog.csdn.net/jz1570551401/article/details/78207208
1. 預安裝，其中libgoogle-glog-dev 可能需要再單獨安裝
sudo apt-get install liblapack-dev libsuitesparse-dev libcxsparse3.1.2 libgflags-dev
libgoogle-glog-dev libgtest-dev
2. 下載ceres：https://github.com/ceres-solver/ceres-solver
3. 解壓文件後進入ceres-master:
mkdir build
cd build
cmake .. //如果這一步驟報錯，可以檢查一下是具體那個庫沒有安裝成功，再sudo apt-get 對應的庫就好
make //大概等幾分鐘
sudo make install
安裝完成後/usr/local/include/ceres 和 /usr/local/lib/ 中
3.2 下載code_utils： https://github.com/gaowenliang/code_utils
解壓至catkin下的src文件，然後在code_utils下面找到sumpixel_test.cpp，修改#include "backward.hpp"爲 #include “code_utils/backward.hpp”，然後運行catkin_make
錯誤提示：code_utils/backward.hpp:216:30: fatal error: elfutils/libdw.h: No such file or directory
compilation terminated.
運行： apt-get install libdw-dev 安裝對應的庫即可，一般安裝這個就可以，如果這樣還不可以試試：sudo apt-get install elfutils
3.3 下載imu_utils: https://github.com/gaowenliang/imu_utils
再次catkin_make
3.4 接下來就可以就可以標定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= "$(find imu_utils)/data/d435i"/>
<param name="max_time_min" type="int" value= "30"/> #標定的時長
<param name="max_cluster" type="int" value= "100"/>
</node>
</launch>
將該launch文件保存成imu.launch至src和install下的launch中， 然後在兩個終端中同時運行
~/catkin_ws_rs$ roslaunch realsense2_camera rs_camera.launch
~/catkin_ws_rs/install/share/realsense2_camera/launch$ roslaunch imu.launch
運行imu.launch文件的終端中會出現
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: 100
gyr x num of Cluster 100
gyr y num of Cluster 100
gyr z num of Cluster 100
acc x num of Cluster 100
acc y num of Cluster 100
acc z num of Cluster 100
wait for imu data.
接下來等待30分鐘即可，標定完成後imu的參數將會被保存在data_save_path的路徑中；
也可以錄製bag保存imu數據後進行標定；https://blog.csdn.net/fang794735225/article/details/92804030
將IMU的標定結果保存到文件imu.yaml中：
#Accelerometers
accelerometer_noise_density: 2.4753056458491202e-02 #Noise density (continuous-time)
accelerometer_random_walk: 5.1092097834339175e-04 #Bias random walk
#Gyroscopes
gyroscope_noise_density: 4.0642638745600332e-03 #Noise density (continuous-time)
gyroscope_random_walk: 2.6537322064011450e-05 #Bias random walk
rostopic: /imu0 #the IMU ROS topic
update_rate: 200.0 #Hz (for discretization of the values above)
相機標定：https://blog.csdn.net/heyijia0327/article/details/43538695
標定realsense的RGB相機，如果使用的是其他相機可以用opencv採集圖像後進行單目標定，這裏不多寫，主要是標定Realsense的彩色相機：
設定好相機的相關參數後(主要是size 和fps)：目前設置圖像大小爲640*480 幀率爲30
roslaunch realsense2_camera rs_camera.launch
啓動realsense相機後，可以用rostopic list 查看對應的打開的topics
可以用ros自帶的標定方法進行相機標定：
~/catkin_ws_rs$ rosrun camera_calibration cameracalibrator.py --size 8x6 --square 0.035 image:=/camera/color/image_raw
最開始的CALBRATE按鈕爲灰色，當採集的圖像符合條件夠就變成綠色，然後可以點擊CALIBRATE進行標定，標定結果顯示如下：
('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])
將標定的結果保存到.yaml文件中 如下所示：
cam0: #640*480
camera_model: pinhole
intrinsics: [653.7064773464425, 654.2058371422769,314.4404941629716, 261.91335661459755]
distortion_model: radtan
distortion_coeffs: [0.20848791559812113, -0.36507062055202405, 0.012079145675854016, -0.006358054188595459]
rostopic: /camera/color/image_raw
resolution: [640, 480]
準備標定板文件aprilgrid.yaml：
相機+IMU標定： https://github.com/ethz-asl/kalibr/wiki/Camera-IMU-calibration#2-collect-images
1. 啓動realsense相機和imu，所以也要先設置好rs_camera.launch文件中的相關參數；
target_type: 'aprilgrid' #gridtype
tagCols: 6 #number of apriltags
tagRows: 6 #number of apriltags
tagSize: 0.021383 #size of apriltag, edge to edge [m]
tagSpacing: 0.2853 #ratio of space between tags to tagSize 0.61cm/2.1383cm = 0.285273348
#example: tagSize=2m, spacing=0.5m --> tagSpacing=spacing/tagSize = 0.25[-]
2. 爲相機和IMU錄製數據包，
可以先運行 rosrun image_view image_view image:=/camera/color/image_raw 查看圖像， 保證在錄製數據包的時候標定板不會超出畫面
錄製的時候儘量保證圖像清晰，不要劇烈移動，採集數據的起始和結束階段注意別晃動太大，如從桌子上拿起或者放下。如果有這樣的動作，在標定階段應該跳過bag數據集的首尾的數據. 同時儘可能地激活IMU的各個角度，各個方向，可以先依次繞各個軸運動，運動完後來個在空中畫8字之類的操作，當然也要注意別運動太劇烈，圖像都模糊了。：
rosbag record /camera/color/image_raw /camera/imu -O rgb_imu.bag
3. 進行標定：參數分別是包名稱(不用加.bag)，相機參數文件，imu參數文件， 標定板文件
~/Project/kalibr_workspace$ kalibr_calibrate_imu_camera --bag ~/Project/kalibr_workspace/rgb_imu.bag.bag --cam ~/Project/kalibr_workspace/camchain.yaml --imu ~/Project/kalibr_workspace/imu.yaml --target ~/Project/kalibr_workspace/aprilgrid.yaml
出現錯誤提示：
importing libraries
Traceback (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 sens
File "/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.optimize
ImportError: No module named scipy.optimize
安裝python相關庫：
sudo apt-get install python-scipy
標定結果如下：
After Optimization (Results)
==================
Normalized Residuals
----------------------------
Reprojection error (cam0): mean 0.552748299525, median 0.473100512902, std: 0.3640429471
Gyroscope error (imu0): mean 0.215478847361, median 0.161409013393, std: 0.19024901177
Accelerometer error (imu0): mean 0.165613674102, median 0.132037510374, std: 0.129116309678
Residuals
----------------------------
Reprojection error (cam0) [px]: mean 0.552748299525, median 0.473100512902, std: 0.3640429471
Gyroscope error (imu0) [rad/s]: mean 0.0123851576362, median 0.00927736573336, std: 0.0109350130176
Accelerometer error (imu0) [m/s^2]: mean 0.0579749018736, median 0.0462211936851, std: 0.04519859501
Transformation 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)
-0.0280180128084
IMU0:
----------------------------
Model: calibrated
Update rate: 200.0
Accelerometer:
Noise density: 0.0247530564585
Noise density (discrete): 0.350061081538
Random walk: 0.000510920978343
Gyroscope:
Noise density: 0.00406426387456
Noise density (discrete): 0.0574773709247
Random walk: 2.6537322064e-05
T_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.yaml
Saving imu calibration to file: imu-homeshinanProjectkalibr_workspacecam_imu640_480_retrgb_imu.bag.yaml
Detailed results written to file: results-imucam-homeshinanProjectkalibr_workspacecam_imu640_480_retrgb_imu.bag.txt
Generating result report...
Timeshift表示的就是IMU相對於camera的延時；
參考博客：https://blog.csdn.net/mxdsdo09/article/details/83514310