Google's DeepMind launched Gemini Robotics 2. This new family of AI models enhances robotic capabilities.
Its key component is Gemini Robotics ER 2. This "embodied reasoning" model functions as a high-level brain for robots. It enables advanced task planning and real-world understanding.
The new system features major upgrades. These include enhanced video understanding, long-horizon planning, and the orchestration of complex tasks.
A significant new feature allows coordination of multiple robot types. These robots can collaborate on a single task.
Google is making the model available to developers. Developers will access the model through the Gemini API. This aims to accelerate progress in general-purpose robotics.