How to use OpenVINS Library [closed]

asked 2020-06-18 06:43:41 -0500

NotARobot gravatar image

updated 2020-06-18 07:07:47 -0500

So, I finally got ROS, OpenCV, and OpenVINS installed on my rpi3 running ubuntu 16.04. I want to be able to send the video from a standard rpi monochrome camera setup and mpu-9250 through a VIN application and send the estimated position of the camera over wifi to the game service unity. Ive been looking into their 'getting started' page, and the tutorial seems very high level. Forgive me, for I do not work much with rpi's and more so arduinos, so most of this is new to me. Im trying to understand how I can do this, so far I know i need to send data to some subscribed topics the program is looking under, yet I do not know what data I need to send for each topic, what program I would use to send it, how, and what format it needs to be in. I also dont know if that is the only thing I need to do. And after all this, I still do not understand fully on how to launch the program(s) after everything is fully said and done.

I have read the documentation they have posted over and over and I know they mention the subjects I am mentioning but the sections they talk about that pertain to them in question I cannot fully conceptualize. If someone who has knowledge in the field of using github repositories or uses VINS specifically and would like to make a more low-level tutorial for someone like me that would be much appreciated, if not, do link me to one where I could find descriptors more comprehensive than those provided by this github. The documentation all the while seems great for this repository, and it is a shame I cannot fully utilise it. As always, all input is appreciated, thanks. =)

edit retag flag offensive reopen merge delete

Closed for the following reason question is off-topic or not relevant by berak
close date 2020-06-18 11:52:14.068882


...and what does it have to do with OpenCV?

kbarni gravatar imagekbarni ( 2020-06-18 06:47:33 -0500 )edit

Well, the OpenVINS library is related to computer vision because its Visual Inertial Navigation, and has OpenCV as one of its dependencies. So I thought the people of OpenCV would have experience using this type of library and could help me out from a beginners standpoint. And thank you for your quick reply! =)

NotARobot gravatar imageNotARobot ( 2020-06-18 07:05:11 -0500 )edit

looking here -- they're using opencv for both the webcam and the aruco pose detection ;

so, if the job is: "get an aruco pose and send it to unity" -- i'm quite sure you can achieve that without ROS and openvins.

again, an XY, not an opencv problem, sorry.

berak gravatar imageberak ( 2020-06-18 11:51:23 -0500 )edit

Huh, I was pretty sure that they just used aruco t calibrate the camera, and then used SLAM and VINS to map the room and localize the position of the camera. Does the aruco tracker have to be there the whole time? I didnt think so....

NotARobot gravatar imageNotARobot ( 2020-06-18 12:15:26 -0500 )edit

No you can use aruco for VIO in OpenVINS. Camera intrinsic and extrinsic calibration can be done with aruco or regular point features as online calibration, but you beed a good initial guess from hand/measuring or kalibr

etan321 gravatar imageetan321 ( 2020-08-12 08:56:31 -0500 )edit