Hi, please can Kalibr perform multicam extrinstics calibration even on systems with non-overlapping FOVs? Thank you.

Hi Patric, I am implementing openvins on d455. The Imu keeps drifting away even if the camera is in stationary. I recalibrate it and the values are very close to the once that already exists in rs_d455 config folder. Do you have any suggestions on how to rectify this?

I'm a begineer here. Is the procedure similar for d435i camera?

Hi Patrick. I was just wondering, would I be able to use a 9 axis imu with OpenVINS?

So complicated

Hello , if i dont have imu and only stereo cameras ,can i pass the result from the first cameras_calibration to VINS FUSION directly?

ty so much !

Hi, I have question regarding when to assure that I have good calibration or not? I still have too much reprojection error, I recorded multiple times with 30 fps and 752x480 and the video contain some glitching, do u suggest to change any paramerts to enhance the calibration? shud the video be super smooth with no glitches?

Hi Patric, Thank you so much for your great video. At timestamp 37:34 you are mentioning the timeshift cam0 to imu0 that its not 0 and we have to take that into account. I have two questions: 1. By we should take that into account, we should use that as the time_offset in the kalibr_imu_chain.yaml file of our open-vins configuration? 2. If this data is bigger than 0.002 seconds, how it will effect our localization?

And barometer data?

Thanks, it helps a lot! <3

Oi Victor

Great video, but why multiply noise and random walk values by some constant value? Spend a lot of time to get accurate results and just multiply on random numbers it's strange, or there are something behind it?

Hi Patrick, how about your computer configuration ? and I have a Intel® Core™ i7-10700 CPU @ 2.90GHz × 16 computer and run one imu and two cameras calibration using kalibr for more than two hours, Could you give me some tips about how to accelarate?

Hi Patrick, great content ! At timestamp 25:48, you said we dont use the values of random walk and noise directly, but inflate thm by order of 10 times or by 20. Why is it so? Is there any mathematical or empirical significance to it? Or is it just that moving robot might incur more noise and bias for IMU than when calculated for steady IMU?

Thank for your great video!

Brilliant.

Hello Patrick, Thank you for your guidance. I need to ask you about the error which I got in OpenVins demonstration step. I follow every step what you did. I will put the error output to your seen. I use rs-d455 and I stuck here. Thank you in advance. """"" process[ov_msckf-1]: started with pid [28691] overriding node verbosity with value from ROS! Setting printing level to: INFO overriding node max_cameras with value from ROS! overriding node max_cameras with value from ROS! overriding node use_stereo with value from ROS! the sensor cam1 of type [NSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE] was not found... the sensor cam1 of type [St6vectorIdSaIdEE] was not found... the sensor cam1 of type [St6vectorIdSaIdEE] was not found... the sensor cam1 of type [St6vectorIiSaIiEE] was not found... the sensor cam1 of type [N5Eigen6MatrixIdLi4ELi4ELi0ELi4ELi4EEE] was not found... overriding node record_timing_information with value from ROS! overriding node record_timing_filepath with value from ROS! overriding node use_stereo with value from ROS! the sensor cam1 of type [NSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE] was not found... the sensor cam1 of type [St6vectorIdSaIdEE] was not found... the sensor cam1 of type [St6vectorIdSaIdEE] was not found... the sensor cam1 of type [St6vectorIiSaIiEE] was not found... the sensor cam1 of type [N5Eigen6MatrixIdLi4ELi4ELi0ELi4ELi4EEE] was not found... subscribing to IMU: /d455/imu the sensor cam1 of type [NSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE] was not found... subscribing to cam (stereo): /d455/color/image_raw subscribing to cam (stereo): /cam1/image_raw unable to parse all parameters, please fix terminate called after throwing an instance of 'boost::wrapexcept<boost::lock_error>' what(): boost: mutex lock failed in pthread_mutex_lock: Invalid argument ================================================================================REQUIRED process [ov_msckf-1] has died! process has died [pid 28691, exit code -6, cmd /home/oguzhan/catkin_ws_kalibr/devel/lib/ov_msckf/run_subscribe_msckf __name:=ov_msckf __log:=/home/oguzhan/.ros/log/43fbcdba-683f-11ee-ab34-e5d2dfc10612/ov_msckf-1.log]. log file: /home/oguzhan/.ros/log/43fbcdba-683f-11ee-ab34-e5d2dfc10612/ov_msckf-1*.log Initiating shutdown! ================================================================================ [ov_msckf-1] killing on exit shutting down processing monitor... ... shutting down processing monitor complete done """""

Thank you for this tutorial! I have a D435i and am waiting to see if anyone is porting OpenVins to ROS 2... Foxy/Humble or higher... currently in installation hell :' (

Thanks for your video. Since librealsense v2.53.1 is last version that supports T265, i would suggest anyone who follows this video should build librealsense from source(branch R/253).

Can you help me design a basic VIO system in Python / C ++ system from scratch

Amazing video, awesome tutorial. I was able get correct pose with two standalone cameras mounted on a drone using this tutorial. I was wondering if you could make a similar video going a bit deeper on understanding the parameters.

Thank you for this video. but how do you generate : "+-- dataset-dir +-- cam0 │ +-- 1385030208726607500.png │ +-- ... │ \-- 1385030212176607500.png +-- cam1 │ +-- 1385030208726607500.png │ +-- ... │ \-- 1385030212176607500.png \-- imu0.csv"?

Thank you!

Really interesting video, I have a short question, can I use this algorithm with other stereo cameras like ZED2 and azure kinect as well?

Is there a need to calibrate the IMU and camera if I simulate a drone and openvins in ROS gazebo?

Hi Patrick. I am currently taking PhD which involves XR testbed. May I kindly ask how to setup ILLIXR? Thank you

Thank you for your great video. I was wondering if: 1. we can collect the required data for camera calibration (static.bag) and visual inertial calibration (dynamic.bag) into one bag file and record camera topics and IMU topics into one bag file? 2. how you record a very long dataset of the IMU? I would be grateful if you could give a brief explanation like what you showed in the video to collect camera data (You stood in front of the AprilGrid and moved the camera in different directions.)

Thank you for the amazing video.💮

Hi, very interesting video, thank you for demonstrating this technology.

Good!!!!!!!