GO-1: The First Universal Embodied Model—No More “See but Can’t Do”

Zhiyuan Robotics unveils GO-1, the first universal embodied model using ViLLA architecture, transforming robot training with multimodal learning and rapid adaptation.

"AI Disruption" Publication 5000 Subscriptions 20% Discount Offer Link.

---

Zhiyuan Robotics' "good stuff" has been revealed, and it comes with a double surprise: the Vision-Language-Latent-Action (ViLLA) architecture and the universal embodied foundational large model, GO-1.

Robot training has long been hampered by the challenge of gathering high-quality data. On the one hand, there is data on the cognitive dimension: the massive amounts of text and image data available on the internet help robots establish basic cognition and understand what the world is like.