A new package for Julia to handle rotations of coordinate frames

[Ronan Chagas]

Hi guys,

I want to let you know that I have just started to code a new package for Julia.
It is called Rotations.jl (https://github.com/ronisbr/Rotations.jl).

My idea is to add all functions related to rotations of coordinate frames, which is very useful for aerospace applications, robotics, etc.
By now, I have only implemented two functions:

angle2dcm: Convert Euler Angles to Direct Cosine Matrix;
dcm2angle: Convert Direct Cosine Matrix to Euler Angles.

I expect to expand this project to contain a lot of functions related to DCM, Quaternions, Rodriguez parameters, etc.
I have been using MATLAB for a long time and now I'm planning to start a project here at my Institution to test Julia.
This project will need a lot of rotations of coordinate frames and, since I have not found any available package for Julia yet, then I started to code one :)
I would appreciate any comments or suggestions to improve this package.

By the way, how can I make this a registered package of Julia?

Best regards,
Ronan

[Tim Holy]

Nice! For rotations in 3d, there's already a Quaternions package
(https://github.com/forio/Quaternions.jl) that implements some related
transformations. Since "Rotations" and "Quaternions" have overlapping-yet-
distinct functionality, I confess to being a bit unsure how this functionality
should be parceled out.

Just to add to the confusion, I also have some code (currently just one file,
AffineTransforms.jl) that I've not yet gotten around to putting up. It includes
an implementation of Itzhack Y. Bar-Itzhack, "New Method for Extracting the
Quaternion from a Rotation Matrix", Journal of Guidance, Control, and
Dynamics, Vol. 23, No. 6 (2000), pp. 1085-1087. This algorithm lets you take
any 3D sorta-close-to-an-orthogonal-rotation-matrix as an input and robustly
calculate the corresponding angle/axis or unit-quaternion representation.

--Tim

[Patrick O'Leary]

On Wednesday, July 16, 2014 3:52:56 PM UTC-5, Tim wrote:

Just to add to the confusion, I also have some code (currently just one file,
AffineTransforms.jl) that I've not yet gotten around to putting up. It includes
an implementation of Itzhack Y. Bar-Itzhack, "New Method for Extracting the
Quaternion from a Rotation Matrix", Journal of Guidance, Control, and
Dynamics, Vol. 23, No. 6 (2000), pp. 1085-1087. This algorithm lets you take
any 3D sorta-close-to-an-orthogonal-rotation-matrix as an input and robustly
calculate the corresponding angle/axis or unit-quaternion representation.

I'm interested in that algorithm--we've been using the SVD to fix sorta-close-to-an-orthogonal-rotation-matrices following Markley, "Attitude Determination using Vector Observations and the Singular Value Decomposition", Journal of the Astronautical Sciences, Vol. 36, No. 3, 1988, pp. 245-258. If nothing else, your way sounds newer. But it might be faster and/or more accurate for situations where a quaternion is the final destination rather than a DCM.

[Ronan Chagas]

Hi Tim,

Good! I know this algorithm developed by Prof. Bar-Itzhack. In fact, everything on my Ph.D. related to aerospace systems was based on theories proposed by him :)

My next step in Rotations.jl is to add the following conversions:

    _ DCM => Euler axis / Euler angle
    _ Euler axis / Euler angle => DCM

This is exactly what you have done in AffineTransforms.jl. Would you like to contribute to Rotations.jl package? :)

About Quaternions, I took a look at the implementation and it seems to be very good!
I think I will use this class and add the code to convert DCM => Quaternions (since Quaternions => DCM is already coded).
The problem with that approach is that the user will need to install both Rotations.jl and Quaternions.jl.

Best regards,
Ronan

[Tim Holy]

Hi Ronan,

On Thursday, July 17, 2014 05:28:25 AM Ronan Chagas wrote:
> This is exactly what you have done in AffineTransforms.jl. Would you like
> to contribute to Rotations.jl package? :)

That would be fine, but FYI there are a lot of other goodies in
AffineTransforms.jl. I just put it up here:

https://github.com/timholy/AffineTransforms.jl

Perhaps give it a glance and we can decide on a way forward with regards to
making one of these an official package.

--Tim

[Ronan Chagas]

Great!

Thanks Tim, I will look and then we can discuss!

Best regards,
Ronan