Google Deepmind launched Gemini Robotics 2 robotic AI

On 30 July, Google Deepmind launched the Gemini Robotics 2 model. It was described that Gemini Robotics 2 enabled robots to reason each and every action, thus unlocking a wide range of tasks. For example, it allows a man-shaped robot to walk, crouch, stretch and manipulate objects in order to clear a messy room. It can even work with other robots to do things faster. This in-depth intelligence can also operate locally in the equipment while being able to adapt seamlessly to the new robotic body within a few hours. Meanwhile, Google DeepMind has published two other robotic AI models - Gemini Robotics ER 2 and On-Device 2. Gemini Romani Robotics ER 2 is the most powerful model of inference (ER), a visual language model (VLM), which will enable robots to communicate with humans, understand the physical world and plan multi-step missions lasting several minutes. On-Device 2 is the most efficient visual-linguistic-action model (VLA) that optimizes local operation on robotic equipment. The model can now quickly adapt to a new robotic entity through hours of data。