Google Launches Gemini Robotics 2 for Real-Time Humanoid Control
DynamicWebPaige · x · 2026-07-31
Google DeepMind has launched Gemini Robotics 2, its next-generation physical AI model designed to act as a unified brain for various robots.
Key breakthroughs of the model include:
- Whole-body Control: The Vision-Language-Action (VLA) model scales to real-time whole-body humanoid control.
- Advanced Dexterity: Supports complex operations ranging from bi-arm grippers to 22DoF Sharpa hands.
- On-device Cross-embodiment: Capable of running across different robot morphologies directly on the device.
- New SOTA in Spatial Intelligence: The Gemini Robotics ER 2 model sets a new state-of-the-art record in real-world spatial intelligence benchmarks.
More from Embodied
- Google Rollsouts Gemini Robotics ER 2 Model via API — ericjang11 · 2026-07-31
- First Contact Tactile Glove Pre-orders Open with 847 Taxels per Hand — adityaag · 2026-07-31
- Over $500M in US Factory Automation Engineered, $1B in New Robotics Requests — MatthewChang · 2026-07-31
- Peter Diamandis Predicts Humanoid Robots Will Walk on the Moon Within 4-6 Years — PeterDiamandis · 2026-07-31
- Tactile Sensing Emerges as Robotics Frontier with Origami Challenge — chris_j_paxton · 2026-07-31
- Google Unveils Gemini Robotics 2.0: Humanoid Robots Gain Whole-Body Intelligence — Ars Technica AI · 2026-07-31