Google DeepMind Advances Robotics AI with Gemini 2.0 Platform

Google DeepMind has introduced Gemini Robotics 2.0, a significant upgrade to its robotic intelligence system. The platform enables robots to tackle complex tasks, continuously perceive and react to changing conditions, and coordinate with other robots. The release includes three new AI sub-models, with one already accessible to developers. The core aim is to build generalist robots—machines capable of performing virtually any task a human could accomplish. DeepMind researchers refer to this objective as achieving "physical AGI." Rather than being narrowly programmed for specific actions, these robots can be given high-level instructions and execute them autonomously. The flagship component is Gemini Robotics ER 2, an upgraded embodied reasoning model representing a substantial improvement over its 1.6 predecessor. Integrated with the Gemini Live API, developers can now experiment with the platform's expanded capabilities. A key achievement is the ability to control full-body humanoid robots with enhanced precision and dexterity, including those equipped with complex humanoid hands.