GR-3

On July 22, the Seed team of ByteDance announced its latest achievement: a generalizable large-scale model of robot manipulation "Seed GR-3" that supports long sequences of complex manipulation tasks. According to the introduction, GR-3 is a large-scale vision-language-action (VLA) model. It shows good generalization ability to new objects, new environments, and new commands with abstract concepts. In addition, GR-3 supports efficient fine-tuning of a small amount of human trajectory data for fast and cost-effective adaptation to new tasks.GR-3 is useful in dealing with long-period and dexterity tasks (including tasks that require two-handed manipulation and ...