When using the API to control the camera and retrieve point clouds, I noticed a discrepancy between the Z values of the point cloud and the actual height values relative to the ground.
The height value of the pick point in the visual results does not match the actual height of the object
When using the API to capture images and obtain point clouds, the result you receive are in the camera reference frame. They do not represent the distances of objects relative to other reference systems (such as the ground or the robot). Therefore, it’s important not to be misled by other reference points. In the camera reference frame, objects that are farther from the camera will have larger Z values.
In a Mech-Vision project, you can use the “Pose Transformation” step to convert these coordinates into the robot’s reference frame. This step is crucial to accurately interpret the spatial relationships and distances between objects in a different reference frame, such as the robot’s reference frame.