This is research code released so the results in our paper can be reproduced. It gets best-effort maintenance only. Replies may be slow, and new features are unlikely to be taken on.
Open an issue with:
- what you ran and what happened;
- your setup —
./scripts/verify.shprints everything relevant; - whether you are using the released models or your own.
Check the README's Limitations section and docs/GRANDTOUR.md first. Most
problems come from a TensorRT version mismatch, missing model files, or wrong
topic and frame names.
Bug fixes are welcome. For anything larger, open an issue first — we may not be able to maintain it.
Before opening a pull request:
- keep the change focused and match the surrounding style;
- check it still builds (
colcon build) and that the demo indocs/GRANDTOUR.mdstill runs; - run
pre-commit run --all-files; - do not add GPL, LGPL or AGPL dependencies, and say what licence anything new is under;
- if you copy code from another project, keep its attribution and add it to
NOTICE; - do not commit models, datasets or recordings. Models are published on the Hugging Face Hub.
By sending a contribution you confirm you have the right to release it under this repository's licence, the RAI Institute Research License (see LICENSE).