Hello Eduardo,
Thank you for the Interest.
1). I have now researched the issue mentioned, and can see the issue is outdated, so i looked through my internal files and the current files on OpenCV github and they are identical.
2). I might not be completely sure on what you mean here. What I have done so far is to look at my setup (The picture I posted) and roughly estimated the transnational part ( I know the dimensions of my 3D design and camera) then the rotational part where I ended up with the pose vector [0,0,60,0,0,135]. What I understand of your suggestion; Look at each frame pair and try to see if the difference in each frame pair is the same as the pose vector or close to?
3). I don’t know if you have seen the code i posted on the link, it is also a bit messy. I have used the Rodrigues rotation formula from the solvePnP 3x1 rotation vector → 3x3 rotation matrix, both for my camera poses and robot poses. My narrow understanding of the formula is that it is an “easy” and efficient way to convert EAA to Rotation Matrix? If so the only thing i have to be concerned about is the order of the representation of EAA (XYZ or ZYX).
If it is Euler angles there are 12 different conventions. I have looked through the manual of UR5-e series, and they only provide the information of rotation vector as default and they support RPY angles, UR5e_Manual, pp.186. Where i researched further and found a support page, UR_Support_page, where they state:
In Universal Robots, the axis-angle representation is used for robot orientation.
So in my mind i can use Rodrigues?