Need to be 100% clear about what quaternion convention is used about px4-matrix HOT 7 CLOSED

Also, you state that your vector multiplication formula is q.inversed() * q(0, v) * q. The glm library convention uses the opposite I believe, because rotating v(1, 0, 0) by quaternion q(cos(rads(90/2)), 0, sin(rads(90/2)), 0) results in r(0, 0, -1). Which was my assumption, but instead I was getting r(0, 0, 1).

from px4-matrix.

Can we make assumption that rotations are always applied right to left?

ie. q2 * q1 * v means apply q1 to v first, then apply q2 rotation.

This would clear things up and be more consistent with how it is done in other places.

from px4-matrix.

@mkschreder

v_rotated = q^(-1) * [0;v] * q

The description for the rotation formula seems to be wrong. As you stated above this actually applies a transformation rather than a rotation.

from px4-matrix.

I switched to Hamilton quaternion def here: 19554c4

The unit tests all pass with previous and current definition, so no significant change, but I agree it is better to follow the common convention. I'm not sure which reference I pulled that from orginaly.

@mkschreder Please post a new issue if you find anything else, thanks for going through the details.

@mkschreder Quaternion multiplication is associative I believe, so I don't think it matters if we we (q2 * q1)*v or q2 * (q1 * v). The code will execute it as QuatMul(q2, Quat_Mul(q1, v)), effectively always evaluating it as q2 * (q1 * v). Am I following you? Did you mean that v is a vector? I don't believe quat * vector is supported in the library.

from px4-matrix.

Roman, vector part of the resulting quaternion from that transformation is
the rotated vector v.

On 16 Aug 2016 09:57, "Roman Bapst" [email protected] wrote:

@mkschreder https://github.com/mkschreder

v_rotated = q^(-1) * [0;v] * q

The description for the rotation formula seems to be wrong. As you stated
above this actually applies a transformation rather than a rotation.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#28 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/AERqibRYSyXXzVEO7xgH3HI0XGuaI19Sks5qgW1mgaJpZM4JlDnZ
.

from px4-matrix.

It is clockwise, as that seems to be the convention when using hamilton
notation (as opposed to positive angle representing counterclockwise
rotation).

On 16 Aug 2016 10:41, "Martin Schroeder" [email protected]
wrote:

Roman, vector part of the resulting quaternion from that transformation is
the rotated vector v.

On 16 Aug 2016 09:57, "Roman Bapst" [email protected] wrote:

@mkschreder https://github.com/mkschreder

v_rotated = q^(-1) * [0;v] * q

The description for the rotation formula seems to be wrong. As you stated
above this actually applies a transformation rather than a rotation.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#28 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/AERqibRYSyXXzVEO7xgH3HI0XGuaI19Sks5qgW1mgaJpZM4JlDnZ
.

from px4-matrix.

Thanks for fixing this.

from px4-matrix.