@DasFrodo I've tried this balance thing in the past with the tilt proportional to the distance from the center, but that method is too extreme at the edges - tends to eject the sphere.
The way this system works is that there's a target sphere above the balance table. The target has opposite co-ords to the rolling sphere so if
sphere x, z = (150, 100)
target x, z = (-150, -100)
like a mirror image. I added some multiplication to tweak it
The table has aim constraint pointing at the target, the tilt effect being less at the edge - sort of arc tangent like the red curve here
so when the ball moves, the table tilts, which drives the upper universal joints, which in turn drive the lower UJs.