Strange behaviour of vrrotvec. What is [0 0 0] axis?

3 views (last 30 days)
Evgheny on 3 Jul 2012
I have such an example:
v0 = [0 0 -1]; v1 = [0 0 1];
q = vrrotvec(v0, v1);
So, as I understand, v1 is v0 rotated by 180 degrees in x-axis. But I get q = [0 0 0 pi]
Why?

Antti on 10 Jul 2012
Edited: Antti on 10 Jul 2012
Actually the rotation could be 180 degrees about any axis that is in the xy-plane, like the y-axis. So the x-axis is not the only correct answer.
To me it seems that the function is not working well. In this case there are infinite choices of correct axis, so the function should return at least one correct axis, for example the y-axis or the x-axis.
However, the axis matters if you want to do this rotation for a 3D body. For a 3D-body the result is different if you rotate 180 degrees about x-axis, or 180 degrees about y-axis. But from those two vectors, it is not possible to determine, whether you want a rotation about x-axis, or y-axis or some other axis in xy-plane.

Jan Danek on 20 Jul 2012
Hi Evgheny and Antti,
Thanks for pointing at this. We have corrected the behaviour of this function, as of MATLAB R2012b it will provide a valid axis for parallel / anti-parallel input vectors.
vrrotvec function is to determine the VRML rotation that will rotate the first input vector to the second. It is written as to find the minimal angle that will do that 0;pi.
Best regards,
Jan