As means of getting familiar with the Animation Rigging workflow I’ve decided to try an rig a robotic arm. The end result I want to achieve is to have an end effector that would drive the rotations of all the arm parts. Each part is constrained to only one axis, so my initial idea was to have one chain constrain on the whole arm and individual Multi-rotation constraints for each part.

So … not a great success

without the Multi-Rotation constraints the arm behaves as expected.

Just to be clear those positions and rotations are “solvable” and increasing solver iterations doesn’t result in any changes. Am I wrong to use this Multi-Rotation Constraint for this or it just doesn’t work well with the chain constrain and I should dived it into smaller IK chains/solvers?

Indeed, multi rotation would need to be solved the same way ChainIK is in order for your rotation constraints to apply like you’d want without changing the whole hierarchy.

As you suggest, you could approach the problem with multiple TwoBoneIKs, ChainIKs to have a more constrained look to your robotic arm.

Well, I haven’t tried yet to port this approach to my robotic arm, but the solution in the Rigging Playground is a custom script based on this: https://meuse.co.jp/?p=2732
So I’m pretty sure 99% of people will give up just by looking at the amount of matrices and linear transformations …

is flipping from negative to positive values. I’m guessing the conversion form quaternions to euler angles combined with all the sine and cosine functions make it oscillate. Will need more time to unpack this.