depth_registered point cloud in the wrong frame_id

asked 2014-10-24 09:39:00 -0500

mark_vision gravatar image

Hi all, I'm using openni2_camera package with the ASUS Xtion on a legged robot. When I look at my own legs I've noticed that the image does not fit my URDF model, it seems shifted. Actually when I manually translate the rgb_optical frame to make it coincident with the depth_optical frame, the VGA fits perfectly but the point cloud still remains wrong because the points are of course the same. So I think that while the system believes that the registration is from depth to RGB, actually the opposite happens, so if I manage to change the way of registration or the frame_id of the published point cloud, it could work. Unfortunately I did not find any parameter to do neither of the two.

Can someone help me? I have Hydro + 12.04

edit retag flag offensive close merge delete