3 ms·

> These discontinuities are not just an artifact of poor implementation; it can be proven that any representation of 3D rotations using only three values must c

> These discontinuities are not just an artifact of poor implementation; it can be proven that any representation of 3D rotations using only three values must contain discontinuities. This is a bit pedantic - and the blog post actually does clarify this - but the problem isn't that a 3D representation of representations has "discontinuities" as such, it's that it's not orientable in Euclidean 3D space. It is similar to the Klein bottle - the mathematical description is continuous, but any Klein bottle made of real-world glass has to intersect itself. Or likewise that a Mobius strip can be demonstrated in 3D but can't be built in Flatland without a 3D entity doing the copy-pasting. Reality just has the wrong topology to represent the full group of rotations. Hence the discussion about projective space later in the blog post. So adding a fourth gimbal is really tantamount to a correctly-oriented embedding of 3D rotations onto a 4-torus (that is, [0,2pi]^4). Gimbal lock also relates to another issue of continuity, related to the "plate dance."[1] Rotations themselves have a sense of continuity (infinitesimal differences in either the angle or axis of rotation, aka they are Lie groups), but Euler angles fail to respect the equivalent fundamental theorem of calculus: adding a bunch of infinitesimal changes might say you are at the identity rotation according to Euler angles, but in reality you have flipped the meaning of the right-hand rule and the overall state of the system is not at the identity. In a robotics context, the robot's hand might have done a complete rotation, but its arm is twisted without the robot "knowing." I believe robotic arms used to have a serious problem with this, either overrotating and breaking the arm, or swinging dangerously fast in the opposite direction. Using quaternions / a fourth gimbal / etc. there would be a measurable phase or pole indicating the true state of the system, and letting the robot know how to rotate its arm without malfunctioning. So, like the Apollo mission, the need for a fourth dimension to keep track of that stuff - and even the quaternion multiplication structure - comes about pretty naturally without ever thinking about abstract math. [1] https://en.m.wikipedia.org/wiki/Plate_trick https://en.m.wikipedia.org/wiki/Plate_trick

tagrun 2y ago
That statement is also incorrect. 1. "any representation of 3D rotations using only three values" That is not representation, that is parametrization. Euler-angle parametrization sometimes fails because it is not a correct parameterization of SO(3) in general by construction, this is why it sometimes fails (essentially, the three consecutive rotations can sometimes effectively collapse into two for certain set of angles, regardless of how you choose your 3 axes, in which case you can't relate 2 independent parameters back to the 3 independent axis-angle parameters). The correct parametrization of SO(3) is the axis-angle parametrization, which can be represented using quaternions or 3D reals matrices. The "representation", on the other hand, would typically be unit quaternions or 3D orthogonal matrices. 2."it can be proven that any representation of 3D rotations using only three values must contain discontinuities." where is that proof and what discontinuity are you talking about? It sound like he misunderstood what "SO(3) is not simply connected" means. Lie groups are differentiable. 3 parameters are sufficient to represent any 3D rotation. The natural parametrization of all Lie groups, including SO(3), is the axis-angle parametrization, and their elements have the form exp(i θ n.J) where n is a unit vector defining the axis of rotation, θ determines the amount of rotation, and J is a vector of the generators of the corresponding Lie algebra. The "regular" 3D matrix representation in the axis-angle parameterization is obtained with so(3) generators L_x, L_y, L_z in their fundamental representation. Basis quaternions i, j, k (which can be represented by Pauli matrices) obey the same Lie algebra as L_x, L_y, L_z, but the group that it corresponds to (which is SU(2)) is a double cover of SO(3) (up to a sign), so they can still be used for implementing 3D rotations once you pick a sign.

tagrun 2y ago

That statement is also incorrect. 1. "any representation of 3D rotations using only three values" That is not representation, that is parametrization. Euler-angle parametrization sometimes fails because it is not a correct parameterization of SO(3) in general by construction, this is why it sometimes fails (essentially, the three consecutive rotations can sometimes effectively collapse into two for certain set of angles, regardless of how you choose your 3 axes, in which case you can't relate 2 independent parameters back to the 3 independent axis-angle parameters). The correct parametrization of SO(3) is the axis-angle parametrization, which can be represented using quaternions or 3D reals matrices. The "representation", on the other hand, would typically be unit quaternions or 3D orthogonal matrices. 2."it can be proven that any representation of 3D rotations using only three values must contain discontinuities." where is that proof and what discontinuity are you talking about? It sound like he misunderstood what "SO(3) is not simply connected" means. Lie groups are differentiable. 3 parameters are sufficient to represent any 3D rotation. The natural parametrization of all Lie groups, including SO(3), is the axis-angle parametrization, and their elements have the form exp(i θ n.J) where n is a unit vector defining the axis of rotation, θ determines the amount of rotation, and J is a vector of the generators of the corresponding Lie algebra. The "regular" 3D matrix representation in the axis-angle parameterization is obtained with so(3) generators L_x, L_y, L_z in their fundamental representation. Basis quaternions i, j, k (which can be represented by Pauli matrices) obey the same Lie algebra as L_x, L_y, L_z, but the group that it corresponds to (which is SU(2)) is a double cover of SO(3) (up to a sign), so they can still be used for implementing 3D rotations once you pick a sign.