Step “Invert Poses”, as I understand it, relates to the spatial distance between two points. I do not know if it is correct.
Does it mean that to move from point A to point B, we need to determine how much movement is needed along the XYZ axes and how much rotation is required in terms of Euler angles?
Is the resulting data essentially an offset value?
Is it equivalent to establishing a coordinate system with point A and determining the position of point B within this coordinate system? Is this the correct understanding?
The inverse of poses module, or what we commonly refer to as the inverse of poses step, does not represent the distance between two points in the space.
Usually, when we mention a pose, we consider it to be in a specific coordinate system (reference frame).
For example, when a camera captures an image of object A, the obtained pose of A is in the coordinate system of the camera.
If we input pose of A in the camera coordinate system to Step “Invert Poses”, this Step inverts the input pose of A and yields the pose of the camera in a coordinate system where object A serves as the origin.
So, if you follow the ‘invert poses’ step with a pose transformation using the ‘UseCorrespondenceInput’ transformation type, does this mean you are transforming point A into the B coordinate system?
In the figure, if A represents the pose in the robot’s coordinate system, and B also represents the pose in the robot’s coordinate system, then after the processing shown in the figure, what you indeed obtain is A in the B coordinate system.