RoboBrain-X0
A cross-embodiment vision-language-action model trained across heterogeneous robot platforms.
Overview
RoboBrain-X0 is a unified vision-language-action model for cross-embodiment action generation. It supports heterogeneous robot platforms through a shared action-token representation.
I contributed to the data pipeline rather than leading the project. My work focused on processing AgiBot-World and DROID data, including action normalization, tokenization, and structured conversion for large-scale VLA pretraining.