[twanvl]

From the article and video I got the impression that the goal was to give animators control over the robot's motion, while maintaining stability. The animators will not have a good model of the physical robot in their head, and that is what this model is correcting.

[icebraining]

I think _Microft's point is that the robot should be able to perform the original motion without oscillating, rather than having to adjust the motion to compensate for oscillation.

[_Microft]

Yes, take the human spine for example: the double s-shape of the spine makes it very difficult to induce oscillations in it and helps absorbing shocks. And then there is the 'Dancer' robot (first example, ca. 20s into the video) with a straight 'spine' ... wobbling around.

[espadrine]

A subtle constraint to optimize is hardware cost.

I assume Disney chose to separate the constraints by going with simple hardware shapes, yielding a predictable cost and an easy construction. A good hardware optimizer would also need to engineer assembly, which seems nontrivial.

Software costs of optimizing motion for existing hardware are more predictable (~19 minutes of an i7-7700 w/ 32GB, per second of animation).

That being said, they acknowledge in their paper that they could optimize certain properties of the robot, like rod size or joint position.