Main Content

# rotmat

Convert quaternion to rotation matrix

Since R2020b

## Description

example

rotationMatrix = rotmat(quat,rotationType) converts the quaternion, quat, to an equivalent rotation matrix representation.

## Examples

collapse all

Define a quaternion for use in point rotation.

theta = 45;
gamma = 30;
quat = quaternion([0,theta,gamma],'eulerd','ZYX','point')
quat = quaternion
0.8924 +  0.23912i +  0.36964j + 0.099046k

Convert the quaternion to a rotation matrix.

rotationMatrix = rotmat(quat,'point')
rotationMatrix = 3×3

0.7071   -0.0000    0.7071
0.3536    0.8660   -0.3536
-0.6124    0.5000    0.6124

To verify the rotation matrix, directly create two rotation matrices corresponding to the rotations about the y- and x-axes. Multiply the rotation matrices and compare to the output of rotmat.

theta = 45;
gamma = 30;

ry = [cosd(theta)   0           sind(theta) ; ...
0             1           0           ; ...
-sind(theta)   0           cosd(theta)];

rx = [1             0           0           ;      ...
0             cosd(gamma) -sind(gamma) ;     ...
0             sind(gamma) cosd(gamma)];

rotationMatrixVerification = rx*ry
rotationMatrixVerification = 3×3

0.7071         0    0.7071
0.3536    0.8660   -0.3536
-0.6124    0.5000    0.6124

Define a quaternion for use in frame rotation.

theta = 45;
gamma = 30;
quat = quaternion([0,theta,gamma],'eulerd','ZYX','frame')
quat = quaternion
0.8924 +  0.23912i +  0.36964j - 0.099046k

Convert the quaternion to a rotation matrix.

rotationMatrix = rotmat(quat,'frame')
rotationMatrix = 3×3

0.7071   -0.0000   -0.7071
0.3536    0.8660    0.3536
0.6124   -0.5000    0.6124

To verify the rotation matrix, directly create two rotation matrices corresponding to the rotations about the y- and x-axes. Multiply the rotation matrices and compare to the output of rotmat.

theta = 45;
gamma = 30;

ry = [cosd(theta)   0           -sind(theta) ; ...
0             1           0           ; ...
sind(theta)   0           cosd(theta)];

rx = [1             0           0           ;      ...
0             cosd(gamma) sind(gamma) ;     ...
0             -sind(gamma) cosd(gamma)];

rotationMatrixVerification = rx*ry
rotationMatrixVerification = 3×3

0.7071         0   -0.7071
0.3536    0.8660    0.3536
0.6124   -0.5000    0.6124

Create a 3-by-1 normalized quaternion vector.

qVec = normalize(quaternion(randn(3,4)));

Convert the quaternion array to rotation matrices. The pages of rotmatArray correspond to the linear index of qVec.

rotmatArray = rotmat(qVec,'frame');

Assume qVec and rotmatArray correspond to a sequence of rotations. Combine the quaternion rotations into a single representation, then apply the quaternion rotation to arbitrarily initialized Cartesian points.

loc = normalize(randn(1,3));
quat = prod(qVec);
rotateframe(quat,loc)
ans = 1×3

0.9524    0.5297    0.9013

Combine the rotation matrices into a single representation, then apply the rotation matrix to the same initial Cartesian points. Verify the quaternion rotation and rotation matrix result in the same orientation.

totalRotMat = eye(3);
for i = 1:size(rotmatArray,3)
totalRotMat = rotmatArray(:,:,i)*totalRotMat;
end
totalRotMat*loc'
ans = 3×1

0.9524
0.5297
0.9013

## Input Arguments

collapse all

Quaternion to convert, specified as a scalar, vector, matrix, or multidimensional array.

Data Types: quaternion

Type of rotation represented by the rotationMatrix output, specified as 'frame' or 'point'.

Data Types: char | string

## Output Arguments

collapse all

Rotation matrix representation, returned as a 3-by-3 matrix or 3-by-3-by-N multidimensional array.

• If quat is a scalar, rotationMatrix is returned as a 3-by-3 matrix.

• If quat is non-scalar, rotationMatrix is returned as a 3-by-3-by-N multidimensional array, where rotationMatrix(:,:,i) is the rotation matrix corresponding to quat(i).

The data type of the rotation matrix is the same as the underlying data type of quat.

Data Types: single | double

## Algorithms

Given a quaternion of the form

$q=a+bi+cj+dk\text{\hspace{0.17em}},$

the equivalent rotation matrix for frame rotation is defined as

$\left[\begin{array}{ccc}2{a}^{2}-1+2{b}^{2}& 2bc+2ad& 2bd-2ac\\ 2bc-2ad& 2{a}^{2}-1+2{c}^{2}& 2cd+2ab\\ 2bd+2ac& 2cd-2ab& 2{a}^{2}-1+2{d}^{2}\end{array}\right]\text{\hspace{0.17em}}.$

The equivalent rotation matrix for point rotation is the transpose of the frame rotation matrix:

$\left[\begin{array}{ccc}2{a}^{2}-1+2{b}^{2}& 2bc-2ad& 2bd+2ac\\ 2bc+2ad& 2{a}^{2}-1+2{c}^{2}& 2cd-2ab\\ 2bd-2ac& 2cd+2ab& 2{a}^{2}-1+2{d}^{2}\end{array}\right]\text{\hspace{0.17em}}.$

## References

[1] Kuipers, Jack B. Quaternions and Rotation Sequences: A Primer with Applications to Orbits, Aerospace, and Virtual Reality. Princeton, NJ: Princeton University Press, 2007.

## Version History

Introduced in R2020b