Difference between camera_link and camera_optical_link

I see most of the sensors have two frames like a camera has camera_link and again camera_optical_link. So, tf is like base_link to camera_link and then camera_link to camera_optical_link. Similarly, I see velodyne_top_base_link and then velodyne_link. Why do we have two frames? What is the difference?

Asked by zrahman on 2023-07-09 17:03:22 UTC

Comments

Answers

For camera_optical_link see #q395930 (I'm pretty certain camera_optical_link is what's camera_rgb_optical_frame is in that Q&A).

As to camera_link: it's typically the base_link of a camera/depth sensor/sensor model: it's the start of the kinematic chain that describes the structure of the sensor, in TF frames directly, or as encoded in a URDF.

One of its main purposes is to provide an attachment point where you'd connect the model of the camera to the rest of your TF tree (such as a robot's end or EEF link).

Together with camera_optical_link, this would help "solve" the problem of the different orientations of the various frames used by cameras (and similar sensors).

Asked by gvdhoorn on 2023-07-10 01:32:31 UTC