DeepMind Announces Plans to Utilize AI Models for Physical Robots
Google DeepMind Introduces New AI Models for Robotics
Overview of the New AI Models
On Wednesday, Google DeepMind announced the launch of two innovative artificial intelligence models aimed at enhancing robotics functionality. Dubbed Gemini Robotics and Gemini Robotics-ER (Extended Reasoning), these models are part of Gemini 2.0, which Google deems as its “most capable” AI technology to date. Unlike traditional AI, which primarily focuses on generating text and images, Gemini Robotics can issue commands that lead to physical actions—thus bridging the gap between digital and physical worlds.
Partnership with Apptronik
To bring these new capabilities into real-world applications, Google will collaborate with Apptronik, a robotics company based in Texas. Known for its work with organizations such as Nvidia and NASA, Apptronik is set to play a crucial role in developing the next generation of humanoid robots powered by Gemini 2.0. Just recently, Google participated in Apptronik’s funding round, which raised a total of $350 million.
Demonstrations of Robotic Capabilities
In a series of demonstration videos, the robots developed by Apptronik showcased their abilities using the new AI models. These included:
- Plugging items into power strips
- Filling lunchboxes
- Moving plastic vegetables
- Zipping up bags
These tasks were performed in response to spoken commands, exemplifying the robots’ interactive capabilities.
Essential Qualities of AI in Robotics
For AI models to be effective in robotic applications, Google highlighted three essential qualities they must possess:
- General Adaptability: The robots must be able to handle different situations effectively.
- Interactive Communication: They should respond quickly to user instructions and changes in their surroundings.
- Dexterity: Robots need to perform tasks that require precision, similar to human finger and hand movements.
Target Audience for Gemini Robotics-ER
The Gemini Robotics-ER model is aimed at roboticists who can use it as a foundation to create their own applications. Currently, this model is available not just to Apptronik but also to a select group of “trusted testers,” which include companies like Agile Robots, Agility Robotics, Boston Dynamics, and Enchanted Tools.
The Competitive Landscape in Robotics
Google is not the only tech giant interested in integrating AI with robotics. In November, OpenAI announced an investment in Physical Intelligence, a startup focused on developing large-scale AI models to drive robotics innovations. Following this, OpenAI also strengthened its robotics initiatives by hiring a former executive from Meta to lead its efforts in this rapidly evolving field. Similarly, Tesla is entering the humanoid robotics scene with its Optimus robot.
Google’s Vision for Robotics and AI
Google CEO Sundar Pichai recently emphasized that the company views robotics as an ideal platform for testing and implementing AI advancements in real-world scenarios. According to Pichai, these robots equipped with Google’s multimodal AI models will have the ability to adapt to their environments dynamically, allowing for enhanced interactions and functionalities.
As advancements continue in this area, the collaboration between AI and robotics promises exciting developments that may transform how we interact with machines in our daily lives.
