DepthAI/OAK cameras

classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

DepthAI/OAK cameras

FPSychoric
Hi there was a basic support for OAK cameras in the standalone version. I know by this thread https://github.com/introlab/rtabmap/issues/742#issuecomment-967497082 that there was issues with rgb alignment and synchronisation or timestamps.

Im finding lots of issues with this cameras in almost every area, but still your partial support seen in that thread is the best I could get, at least with ROS

I would ask you if you have planned or if it is in the works give full support to this cameras (specially in looking for mobile platform use)
If don't I would ask if is possible the reason, in example closed firmware, lack of support from the company , hardware issues. (I found issues in all that in my case)

I ask you this because I invested some money and everything I have to can be dedicated to robotics (and I have not much for anything) on this cameras and in a few projects with them, and I'm thinking in surrender as I cannot see almost anything positive and in the right way (reliable)  and for me the right way is use your software.

No problem if you decide don't answer, or do it in private, to be honest I'm quite frustrated because every sounds great with DetphAI but later the real thing is very different (in the slam/depth/ROS related)

Cheers
Reply | Threaded
Open this post in threaded view
|

Re: DepthAI/OAK cameras

FPSychoric
Thank you Matlabbe.
Reply | Threaded
Open this post in threaded view
|

Re: DepthAI/OAK cameras

matlabbe
Administrator
Hi,

You were right on time, I was working with oak-d today. See my last posts here: https://github.com/introlab/rtabmap/issues/742#issuecomment-1059442017 and here: https://github.com/introlab/rtabmap/issues/742#issuecomment-1059600320

The second one I added a parameter on rtabmap side to filter bad synchronized frames. Hope those examples will help you!

cheers,
Mathieu

Reply | Threaded
Open this post in threaded view
|

Re: DepthAI/OAK cameras

matlabbe
Administrator
When the IMU urdf issue will be fixed, I will update the official rtabmap hand-held mapping wiki with those oak-d examples.
Reply | Threaded
Open this post in threaded view
|

Re: DepthAI/OAK cameras

FPSychoric
yeah Thanks!!! , I seen it as I was part of the thread, so I was subscribed. I couldn't believe so fast you made it work and how you find issues so fast, thanks a lot.
As you seen there was issues that needed attention, because that I was so frustrated. After that I discovered hardware issues in the two cameras I have access. Also they just applied some improvements to one example, but not to the ones I was using so they weren't working at all. I understood it was your case too.
I will wait to they fix the problem that you said and to your guide and everything updated in github because I'm very bad with patches and that stuff.

I think other issues, (as rational_polynomial instead plumb_bob/equidistant that prevent other slam packages of can use it) but for the moment fixing what you found, will be a lot easier to debug them

Thanks!!


Reply | Threaded
Open this post in threaded view
|

Re: DepthAI/OAK cameras

FPSychoric
In reply to this post by FPSychoric
I'm receiving this error from Rtabmap and main branch of depthai, I guess due to the imu interpolation upgraded in recent branchs. it send me different erros in diffterent modes, 0,1 or 2.
i.e.
`[ERROR] (2022-04-29 12:18:10.218) CoreWrapper.cpp:2506::imuAsyncCallback() IMU received doesn't have orientation set, it is ignored.
Reply | Threaded
Open this post in threaded view
|

Re: DepthAI/OAK cameras

matlabbe
Administrator