topic Re: probable error in static transforms in Items with no label

probable error in static transforms

SMish7 — Mon, 24 Jul 2017 10:22:48 GMT

Hi,

I wanted to know the displacements between the various cameras on the euclid sensor and for getting this information i did a rostopic echo /tf_static on my euclid device's terminal. Although the transform data given for most frames seems correct, I think the transfrom data for the fisheye camera frame is wrong. I say so because camera_fisheye_frame is the child of camera_link frame and the camera_link frame coincides with the camera_rgb_frame. So, the camera_fisheye_frame must be about 8 to 9 cms toward the left of the camera_rgb frame but the information published by the /tf_static topic does not correspond to this. I will be grateful if someone can verify this can give me the correct displacement between the camera_rgb frame and the camera_fisheye_frame.

Re: probable error in static transforms

idata — Mon, 24 Jul 2017 19:37:57 GMT

Hello SubMarIn,

Thanks for reaching out!

There are several known issues with Euclid's fisheye camera as you can see in https://communities.intel.com/docs/DOC-111978 https://communities.intel.com/docs/DOC-111978. However, I don't see any issue regarding a misplacing of the fisheye's frame.

Please let us investigate about this, we'll get back to you as soon as possible.

Pedro M.

Re: probable error in static transforms

SMish7 — Mon, 24 Jul 2017 19:48:30 GMT

Hi,

You can do a rostopic echo /tf_static and check it.

Re: probable error in static transforms

idata — Wed, 09 Aug 2017 18:29:34 GMT

Hello SubMarIn,

We have analyzed the issue that you reported and it indeed looks like a bug. We have submitted it to the Euclid development team for further investigation.

We will update this thread when we have more information.

Pedro M.

Re: probable error in static transforms

Meitav_K_Intel — Thu, 10 Aug 2017 09:02:58 GMT

Hi guys,

I printed out the /tf_static and the values for the fisheye transform that I got were around 2.7-3cm.

The distance between the fisheye and the left IR camera (to which everything is aligned) is 30mm, so the results are accurate.

I hope this makes more sense now.

Meitav

Intel Euclid Development Team