Thanks to the RoboChallenge dataset and the real-world testbed, which are undoubtedly in advancing embodied intelligence.
I’ve noticed a few issues with the RoboChallenge data:
- Because of breaking changes in the LeRobot version, the original conversion scripts does not work.
- I find each data subset is tied to one specific robot configuration, so conversion requires different settings.
Could you please provide a mapping between every dataset and its corresponding robot configuration? This would make it much easier to write the appropriate conversion scripts.