Chessboard pose in robot and camera world

edit

I have a robot holding a chessboard in front of a camera. In know the chessboard's position in the robot world. If I do pose detection of this chessboard I also now it's position in the camera world. Both cooridnates together define the relation between camera world and robot world. This all works okay, but of course, there are some tolerances.

My question is related to the detection of the chessboard pose with the camera. What is the best way to position the chessboard in view of the camera:

1. Perpendicular
2. Rotated around 1,2 or three axes of the camera world
3. Other.

Which will give the most reliable results?