Gimbal Lock Presentation Jan 2016

Rotations in 3D Graphics and the Gimbal Lock
Valentin Koch
Autodesk Inc.
January 27, 2016
Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 1 / 37

Presentation Road Map
1 Introduction
2 Rotation Matrices in R2
4 Gimbal Lock
5 Quaternions

About me
MSc. Mathematics with thesis

in Mathematical Optimization
Principal Research Engineer at
Autodesk, Inc.
Infrastructure Optimization
3D environment, InfraWorks
(similar to Sim City).
Encountered Gimbal Lock using
numerical optimization
algorithms.

1 Introduction
4 Gimbal Lock
5 Quaternions

Rotation Matrix in R2
Rotation on the unit circle
A vector (x, y ) of magnitude r , is rotated by an angle t about the origin.

We recall that
x⇤
cos t = ,
r
y⇤
sin t = ,
r
and if r = 1, we obtain
x ⇤ = cos t,
y ⇤ = sin t.

Rotating an arbitrary point
Given a point (x, y ) at an angle ↵ on the unit circle. We want to rotate it

by angle . Hence
x ⇤ = cos(↵ + ).
Using the trig identities, we see that
x ⇤ = cos(↵ + ) = cos ↵ cos sin ↵ sin

= x cos y sin .
Similarly, we have
y ⇤ = y cos + x sin .

Previous result in Matrix notation
Since
x ⇤ = x cos y sin ,
⇤
y = y cos + x sin ,
we write ✓ ◆ ✓ ◆✓ ◆
x⇤ cos sin x
=
y⇤ sin cos y
and our rotation matrix in R2 is
✓ ◆
cos sin
R= .
sin cos

1 Introduction
4 Gimbal Lock
5 Quaternions

Rotation Matrices in R3
Major axis approach
How to extend previous result to R3 ? General idea:

R2 rotation around major axis x, y , and z
Extend the 2x2 matrix to 3x3
Question
Why selecting 3 axis to rotate an object in R3 ?

Example rotation about z-axis
Example
Step 1 Take 2x2 rotation matrix
✓ ◆
cos sin
R=
sin cos
.
Step 2 Extend to 3x3 by adding z-axis, and keep z value unchanged
0 1
cos sin 0
Rz = @ sin cos 0A .
0 0 1

Basic rotation matrices
Given angles ↵, , and , we obtain the basic rotations about the x-axis
0 1
1 0 0
Rx = @0 cos ↵ sin↵A ,
0 sin ↵ cos ↵
the y-axis 0 1
cos 0 sin
Ry = @ 0 1 0 A,
sin 0 cos
and the z-axis 0 1
cos sin 0
Rz = @ sin cos 0A .
0 0 1

Disadvantages
Why are basic rotation matrices bad?

Expensive to correct matrix drifting.
Hard to interpolate nicely between two rotations.
Gimbal Lock!

Matrix drift
Matrix drift happens when multiple matrices are concatenated.

Round o↵ errors happens on some of the matrix elements, resulting in
sheared rotations.
Need to orthonormalize the matrix.
Gram-Schmidt process is computationally expensive!

Linear Interpolation (LERP))
Example
⇡
Linearly interpolate between identity I and rotation A, which is 2 around
x-axis.
0 1 0 1 0 1
1 0 0 1 0 0 1 0 0
R = 0.5 @0 1 0A + 0.5 @0 0 1 A = @0 0.5 0.5A
0 0 1 0 1 0 0 0.5 0.5
Let u = (0, 1, 0)T . Then kIuk = kAuk = 1.

p
But kRuk = k(0, 0.5, 0.5)T k = 0.5.
So R is not a rotation matrix!

You need a Spherical Linear Interpolation (SLERP) for the rotational
parts, and LERP for the other parts. Complicated.
1 Introduction
4 Gimbal Lock
5 Quaternions

Gimbal Lock
Sword movement
We want to rotate a sword by = ⇡ about the y -axis, and use rotations

about the z and y -axis to move the sword down and up again.

Gimbal Lock
Compose rotation matrices
We first want to rotate to = ⇡2 . Let R = Rx Ry Rz , which is

0 10 10 1
1 0 0 cos 0 sin cos sin 0
@
R = 0 cos ↵ sin ↵ A @ 0 1 0 A @ sin cos 0A .
0 sin ↵ cos ↵ sin 0 cos 0 0 1
Since cos ⇡2 = 0, and sin ⇡2 = 1, the above becomes
0 10 10 1
1 0 0 0 0 1 cos sin 0
R = @0 cos ↵ sin ↵A @0 1 0A @ sin cos 0A ,
0 sin ↵ cos ↵ 1 0 0 0 0 1
which equals
0 1
0 0 1
@
R = sin ↵ cos + cos ↵ sin sin ↵ sin + cos ↵ cos 0A .
cos↵ cos + sin ↵ sin cos ↵ sin + sin ↵ cos 0

Gimbal Lock
Simplification of rotation matrix
Using the facts that
sin(↵ ± ) = sin ↵ cos ± cos ↵ sin

cos(↵ ± ) = cos ↵ cos ⌥ sin ↵ cos
we obtain
0 1
0 0 1
R = @ sin(↵ + ) cos(↵ + ) 0A
cos(↵ + ) sin(↵ + ) 0

Gimbal Lock
Loosing a degree of freedom
From = 0 to ⇡2 , we now want to rotate 6⇡ about the z-axis, and from

= ⇡2 to ⇡, we want to rotate 6⇡ about the x-axis. But since
0 1
0 0 1
R = @ sin(↵ + ) cos(↵ + ) 0A ,
cos(↵ + ) sin(↵ + ) 0
is a rotation matrix about the z-axis, changing ↵ or has the same

e↵ect! The angle ↵ + may change, but the rotation happens always
about the z-axis with unexpected results.
Watch this Video by PuppetMaster’s 3D experiences, CC-BY.

Gimbal Lock
Euler angles
Using Euler angles to steer pitch, roll, and yaw.
Pictures by MathsPoetry, CC BY-SA

Rotation Alternatives
How to avoid Gimbal Lock?
There are alternative rotations:

Euler angles
Axis-angle representation
Quaternions

Axis angle representation
Euler vector
A rotation vector
r = ✓ê,
where ê = (ex , ey , ez )T is an
arbitrary unit vector, and ✓ is the
angle of rotation about the axis ê.
No Gimbal Lock.
Combining two rotations defined by Euler vectors is not simple.
Cannot use LERP to interpolate.

1 Introduction
4 Gimbal Lock
5 Quaternions

Quaternions
The holy grail of 3D rotations
Introduced by William Rowan

Hamilton in 1843
An extension to complex
numbers from two dimensions to
three
i 2 = j 2 = k 2 = ijk = 1
by JP, CC BY-SA 2.0

Quaternions
Representation
Definition
A quaternion is a quadruple formed by a scalar w , and a vector
v = (x, y , z). We write it as
q = (w , v).

Quaternions
Unit Quaternions
Any rotation in R3 can be represented by a unit quaternion.
Definition
A unit quaternion is a quaternion q, such that kqk = 1, where
kqk = w 2 + x 2 + y 2 + z 2 .
All unit quaternions form a hypersphere in R4 . Rotations happen on the

surface of this hypersphere.

Quaternions
From axis-angle to quaternions
Given an axis defined by unit vector
u = (ux , uy , uz ),
and an angle
✓.
How do I obtain a rotation quaternion?

Quaternions
Complex numbers and R2
Any point in R2 can be represented by a complex number, where the real

part lays on the x-axis, and the imaginary part on the y -axis.
Euler’s formula relates the trigonometric functions to the exponential

function
by gunther and Wereon, CC BY-SA 3.0

Quaternions
Extending Euler’s Formula
Let i = (1, 0, 0)T , j = (0, 1, 0) and k = (0, 0, 1). Given u and ✓, an

extension to Euler’s Formula says that
✓ ✓ ✓
ew v = e 2 (ux i+uy j+uz k) = cos + sin (ux i + uy j + uz k).
2 2
Conversion of an axis-angle to a rotation quaternion

Given a unit vector u = (ux , uy , uz ) 2 R3 and a rotation angle ✓ about u,
the corresponding unit quaternion q is
✓ ✓
q = (cos , sin u).
2 2

Quaternions
Axis-angle to quaternion example
Example
We want to rotate 90 about the y -axis using a quaternion.
We know q = (cos 2✓ , sin 2✓ u). The y -axis can be represented by

u = (0, 1, 0), and ✓ = ⇡2 . We obtain the unit quaternion
⇡ ⇡
q = (cos , 0, sin , 0).
4 4
Great. What do we do with this rotation quaternion now?

Quaternions
Rotation
Rotation
Any vector v 2 R3 can be rotated using a rotation quaternion q by
1
qpq ,
where p = (0, v), using the Hamilton product.
What is the Hamilton product and how do you take the inverse of a
quaternion?

Hamilton Product
Definition
Let q0 = (w0 , v0 ) and q1 = (w1 , v1 ). The Hamilton Product is defined as
q0 q1 = (w0 w1 v0 · v1 , w0 v1 + w1 v0 + v0 ⇥ v1 ).
Using transposes q0 = (w0 , x0 , y0 , z0 )T and q1 = (w1 , x1 , y1 , z1 )T , we can

write 0 1
w 0 w 1 x0 x 1 y 0 y1 z 0 z 1
Bw 0 x 1 + x 0 w 1 + y 0 z 1 z 0 y 1 C
q0 q1 = B C
@w 0 y 1 x 0 z 1 + y 0 w 1 + z 0 x 1 A .
w 0 z 1 + x0 y1 y0 x1 + z 0 w 1
Warning
Quaternion multiplication is not commutative, q0 q1 6= q1 q0

Quaternion Inverse
Definition
The inverse of a quaternion q = (w , v) is computed as
1 (w , v)
q = .
kqk2
Addition and subtraction is the same as with complex numbers.

Quaternions
Rotation matrix
Given a rotation quaternion q = (w , x, y , z), we can rotate any vector

v 2 R3 using the
Rotation matrix
0 1
2y 2 2z 2
1 2xy 2zw 2xz + 2yw
Rq = @ 2xy + 2zw 1 2x 2 2z 2 2yz 2xw A .
2xz 2yw 2yz + 2xw 1 2x 2 2y 2

Quaternions
Compose rotations
Given a rotation defined by q0 , followed by q1 . They can be composed as

q⇤ = q1 q0 .
Example
Sword movement We rotate = ⇡2 about the y -axis v0 = (0, 1, 0), followed
by ↵ = 6⇡ about the x-axis v1 = (1, 0, 0). We obtain the quaternions
⇡ ⇡ ⇡ ⇡
q0 = (cos , 0, sin , 0), q1 = (cos , sin , 0, 0).
4 4 12 12
Compose them q = q1 q0 = (w , x, y , z), we get
⇡ ⇡ ⇡ ⇡
w = cos cos , x = sin cos ,
12 4 12 4
⇡ ⇡ ⇡ ⇡
y = cos sin , z = sin sin .
12 4 12 4
Quaternions
Interpolation
The same as with complex numbers.

LERP
qt = (1 t)q0 + tq1

References and Links
Links:
Gimbal Lock
Rotating Objects Using Quaternions
Understanding Quaternions

Gimbal Lock Presentation Jan 2016

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Gimbal Lock Presentation Jan 2016

Uploaded by

Copyright:

Available Formats

Rotations in 3D Graphics and the Gimbal Lock

January 27, 2016

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 1 / 37

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 2 / 37

MSc. Mathematics with thesis

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 3 / 37

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 4 / 37

A vector (x, y ) of magnitude r , is rotated by an angle t about the origin.

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 5 / 37

Given a point (x, y ) at an angle ↵ on the unit circle. We want to rotate it

x ⇤ = cos(↵ + ) = cos ↵ cos sin ↵ sin

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 6 / 37

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 7 / 37

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 8 / 37

How to extend previous result to R3 ? General idea:

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 9 / 37

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 10 / 37

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 11 / 37

Why are basic rotation matrices bad?

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 12 / 37

Matrix drift happens when multiple matrices are concatenated.

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 13 / 37

Let u = (0, 1, 0)T . Then kIuk = kAuk = 1.

So R is not a rotation matrix!

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 15 / 37

We want to rotate a sword by = ⇡ about the y -axis, and use rotations

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 16 / 37

We first want to rotate to = ⇡2 . Let R = Rx Ry Rz , which is

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 17 / 37

Using the facts that

sin(↵ ± ) = sin ↵ cos ± cos ↵ sin

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 18 / 37

From = 0 to ⇡2 , we now want to rotate 6⇡ about the z-axis, and from

is a rotation matrix about the z-axis, changing ↵ or has the same

Watch this Video by PuppetMaster’s 3D experiences, CC-BY.

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 19 / 37

Using Euler angles to steer pitch, roll, and yaw.

Pictures by MathsPoetry, CC BY-SA

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 20 / 37

How to avoid Gimbal Lock?

There are alternative rotations:

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 21 / 37

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 22 / 37

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 23 / 37

Introduced by William Rowan

by JP, CC BY-SA 2.0

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 24 / 37

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 25 / 37

Any rotation in R3 can be represented by a unit quaternion.

All unit quaternions form a hypersphere in R4 . Rotations happen on the

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 26 / 37

Given an axis defined by unit vector

How do I obtain a rotation quaternion?

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 27 / 37

Any point in R2 can be represented by a complex number, where the real

Euler’s formula relates the trigonometric functions to the exponential

by gunther and Wereon, CC BY-SA 3.0

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 28 / 37

Let i = (1, 0, 0)T , j = (0, 1, 0) and k = (0, 0, 1). Given u and ✓, an

Conversion of an axis-angle to a rotation quaternion

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 29 / 37

We know q = (cos 2✓ , sin 2✓ u). The y -axis can be represented by

Valentin Koch (ADSK) IEEE Okanagan January 27, 2016 30 / 37

where p = (0, v), using the Hamilton product.