DeepMind Announces Plans to Utilize AI Models for Physical Robots
Google DeepMind Launches Advanced AI Models for Robotics
On Wednesday, Google DeepMind unveiled two new artificial intelligence models tailored for robotic applications. Both models operate on Gemini 2.0, which Google has described as its most advanced AI framework to date.
Partnership with Apptronik
To enhance its robotics capabilities, Google announced a collaboration with Apptronik, a robotics company based in Texas. This partnership aims to develop the next generation of humanoid robots using the Gemini 2.0 technology. Apptronik has an impressive history, having collaborated with notable companies like Nvidia and agencies such as NASA. Recently, they completed a $350 million funding round, which included participation from Google.
Demonstration of Robotic Abilities
In recent demonstration videos, Google showcased Apptronik robots equipped with the new AI models performing various tasks. These robots were seen plugging devices into power strips, packing lunchboxes, moving plastic vegetables, and zipping bags—all driven by verbal commands. Although this technology is exciting, Google has not provided specific timelines for its commercial release.
Key Qualities of AI in Robotics
In a blog post detailing the new models, Google emphasized three essential qualities AI systems must possess to be effective in robotics:
General Adaptability: Robots must be capable of adjusting to different scenarios and challenges they encounter in real time.
Interactive Response: The AI needs to quickly understand commands and respond to changes in their environment efficiently.
- Dexterity: It’s crucial for robots to possess the hand-like manipulation skills necessary to perform tasks that humans would typically handle, such as delicately moving or organizing objects.
Gemini Robotics-ER: Customizable AI for Robotics
The second model, Gemini Robotics-ER (Extended Reasoning), aims to serve as a foundation for roboticists looking to create their own AI models. This version is currently available to trusted partners like Agile Robots, Boston Dynamics, and Enchanted Tools, allowing them to explore innovative applications in robotics.
Growing Interest in AI Robotics
Google is not alone in its quest to integrate AI technology into the realm of robotics. In November, OpenAI made headlines by investing in Physical Intelligence, a startup focused on developing general-purpose AI models to be utilized in physical robotics. This initiative underscores a growing trend of AI companies venturing into robotic solutions.
Additionally, OpenAI recently hired an executive from Meta who previously led the company’s augmented reality project to spearhead its robotics efforts. Similarly, companies like Tesla are also actively entering the humanoid robotics market, as evidenced by their work on the Optimus robot.
Google’s Vision for AI in Robotics
Sundar Pichai, CEO of Google, shared insights on social media regarding the company’s ambitions, stating that robotics provides a valuable platform for testing AI advancements in practical, physical applications. He highlighted that the robots would utilize Google’s multimodal AI models, which allow for flexibility and adaptability in their operations.
With these developments, Google and its partners are poised to push the boundaries of robotics, making significant strides in how AI can enhance physical tasks and improve interaction in everyday settings.
