Google Deepmind launches GeminiRobotics2 robot AI model

On July 30, Google Deepmind launched the GeminiRobotics2 model. According to reports, GeminiRobotics2 enables robots to reason for every action, unlocking a wide range of tasks. For example, it could allow a humanoid robot to walk, squat, stretch and manipulate objects to clean up a cluttered room. It can even work with other robots to get work done faster. This deep intelligence can also run locally on the device while seamlessly adapting to the new robot body in just a few hours. At the same time, Google DeepMind also released two other robot AI models-GeminiRobotics ER2 and On-Device2. GeminiRobotics ER2 is the most powerful embodied reasoning (ER) model, a visual language model (VLM) that will enable robots to communicate with humans, understand the physical world, and plan multi-step tasks that last minutes. On-Device2 is the most efficient visual-language-action model (VLA), optimized to run locally on robotic equipment. The model can now quickly adapt to new robotic entities using hours of data.