Google is integrating its DeepMind artificial intelligence models into the real world to enhance robotics capabilities. On Wednesday, the company introduced two new AI models: Gemini Robotics and Gemini Robotics-ER (extended reasoning), both powered by Gemini 2. 0, which Google describes as its "most capable" AI to date. Gemini Robotics extends beyond traditional generative AI outputs like text and images, enabling robots to respond to physical action commands. In a blog post, Google announced a partnership with Apptronik, a Texas-based robotics firm, to create the next wave of humanoid robots utilizing Gemini 2. 0. Apptronik has collaborated with Nvidia and NASA in the past, and last month, the company confirmed that Google participated in its $350 million funding round. In demonstration videos, Google showcased Apptronik robots using the new AI models to perform tasks such as plugging devices into power strips, packing lunchboxes, moving plastic vegetables, and zipping bags, all in response to voice commands.

However, the company did not specify when this technology will become commercially available. Google emphasized that for AI models to be effective in robotics, they must possess three key qualities: they need to be general to adapt to various situations, interactive to quickly understand and respond to instructions or environmental changes, and dexterous to perform tasks similar to what people do with their hands. Gemini Robotics-ER is specifically intended for roboticists to use as a framework for training their own models. It is accessible to Apptronik and a select group of "trusted testers, " including Agile Robots, Agility Robotics, Boston Dynamics, and Enchanted Tools. Google is not alone in its efforts to apply AI in robotics. In November, OpenAI invested in Physical Intelligence, a startup aiming to "bring general-purpose AI into the physical world" through the development of large-scale AI models and algorithms that empower robots. In conjunction with that investment announcement, OpenAI appointed the former head of Meta's Orion augmented reality glasses project to direct the startup's initiatives in robotics and consumer hardware. Tesla is also venturing into the rapidly advancing humanoid robotics sector with its Optimus robot. On Wednesday, Google CEO Sundar Pichai shared on X that the company views "robotics as a valuable testing ground for applying AI advancements in the physical realm. " Pichai indicated that these robots will leverage Google's multimodal AI models to "adapt to their surroundings and make adjustments in real time. "