Brainstorm missing features

[mmattamala]

Hi @JonasFrey96 @YifuTao

This and next week I plan to migrate the alphasense_rsl repo currently on Bitbucket to this repo, and make it open source. In that way we can make the repo easily accessible by collaborators (like DRS in Oxford) and managing the project with GitHub. My main concern is keeping this also synchronized with the changes in anymal_rsl

In particular, some changes I have in mind:

• Enable options to publish original image, undistorted, and mono
• Publish undistorted + mask of valid pixels, so we have access to the full undistorted image without cropping
• Improve auto white balance model
• Improve color calibration script (including changes in cpp code and python bindings to accept the bias term)

Feel free to contribute with any other ideas here in the meantime, and then we can make separate issues for that once I have a better idea on how to organize this. @JonasFrey96 also feel free to add other people might be interested or are using the alphasense at RSL, like Turcan or Fan.

[mmattamala]

Other idea: Enabling options to disable publication of compressed and theora images at the roslaunch level (to reduce the CPU load)

[mmattamala]

Implement support for CPU only in ffcc_catkin (so we can do auto white balance on the CPU too)

[nubertj]

Hi guys, this is not an rsl version of an alphasense driver, right? If it is really stuff from you I would recommend calling it rsl_alphasense instead, to be inline with the other packages we have in our github.

[mmattamala]

Thanks for the comment. This will be the same repo that currently exists in Bitbucket with the same name. It's not a driver, it's the debayer code.

I think that changing the name to follow the conventions make sense. In any case we need to ensure that the changes are propagated to anymal_rsl and other related repos.

[mmattamala]

Closing as most of the comments were addressed in #2:

• Enable options to publish original image, undistorted, and mono
  • -> A solution was discussed and agreed with @lintong-zhang . The README now explains what's published
• Publish undistorted + mask of valid pixels, so we have access to the full undistorted image without cropping
  • -> Discarded for now, we can mask on the client side
• Improve auto white balance model
  • -> Seems to be working fine, color calibration also seems to provide sensible results
• Improve color calibration script (including changes in cpp code and python bindings to accept the bias term)
  • -> This was addressed with @YifuTao on the Python side, and also integrated on the cpp side in #2
• Enabling options to disable publication of compressed and theora images at the roslaunch level (to reduce the CPU load)
  • -> This was added as an option to the default roslaunch as an example