GSOC 2026 Ideas

GSOC 2026 Proposal: RoboStack - Robotics Software and GPU on Conda

The RoboStack project wants to grow to support more robotics projects and provide a better experience for users of the ROS ecosystem. This proposal lists ideas on how to improve RoboStack and what related focus areas could be worked on during Google Summer of Code 2026.

1. Better GPU Support

Estimate: 175h
Mentors: Wolf Vollprecht, Ruben Arts

RoboStack doesn't build any GPU enabled packages yet. Adding GPU support to popular robotics libraries like OpenCV, PCL, and PyTorch (Jetson) would be a significant improvement. This would involve building and testing these packages with GPU support on Nvidia (Jetson) devices, or potentionally also AMD ROCm cards.

• Opencv: current recipe CUDA libs
• PCL: current recipe PCL CUDA support

Note: to do this task, we would modify the recipes on conda-forge and find potential other areas of GPU-improvement for ROS packages themselves.

2. Support for New ROS Distributions

Estimate: 350h
Mentors: Wolf Vollprecht, Ruben Arts

RoboStack currently supports noetic, foxy, humble, jazzy, kilted. Adding support for newer ROS distributions like rolling would help keep the project up to date and provide users with access to the latest features and improvements in the ROS ecosystem. To support rolling, the contributor would need to help evolve the current build system to better handle rolling releases. As the effort would increase with a faster moving distribution.

This could involve the use of AI tools to help upstream or reapply patches to packages, inspect build failures and help with local debugging of the packages.

3. Improved Documentation and Tutorials

Estimate: 175h
Mentors: Wolf Vollprecht, Ruben Arts

There is lots of room to help users with improved tutorials and documentation. This could include step-by-step guides on how to install and use RoboStack with different ROS distributions, as well as tutorials on how to build and contribute new packages to the RoboStack ecosystem.

• Tutorials:
  • (Improving) How to install RoboStack channels with different ROS distributions
  • (Improving) How to build and contribute new packages to RoboStack
  • How to use RoboStack with popular robotics libraries like OpenCV, PCL, and PyTorch
• Documentation:
  • Detailed documentation on the build process and package management in RoboStack
  • FAQs and troubleshooting guides for common issues faced by users

Disclaimer

This issue is used in the GSOC 2026 application.

Start with the linked conda-forge OpenCV and PCL recipe build.sh files, then review the listed ROS distribution repositories and existing RoboStack documentation. Choose one proposal and agree on a narrower scope with the mentors before implementation; done should include the defined change, build or test validation, and any required documentation.