DeepMind Announces Plans to Utilize AI Models for Physical Robots
Google DeepMind Launches Advanced AI Robotics Models
Introduction of Gemini 2.0
On Wednesday, Google DeepMind introduced two new artificial intelligence models specifically designed for robotics. Both models, known as Gemini Robotics and Gemini Robotics-ER (Extended Reasoning), operate on what Google describes as its "most capable" AI framework, Gemini 2.0. These new models aim to advance the capabilities of robots beyond generating text and images, enabling them to perform physical tasks in the real world.
Partnership with Apptronik
In a strategic move, Google has partnered with Apptronik, a robotics company based in Texas, to develop the next generation of humanoid robots using Gemini 2.0 technology. Apptronik has a strong background in robotics, collaborating with notable organizations like NASA and Nvidia. Recently, Google also participated in a $350 million funding round for Apptronik, underscoring its commitment to advancing robotic technology.
Practical Demonstrations
Google showcased the capabilities of these new AI models in several demonstration videos. The Apptronik robots were seen executing various tasks such as:
- Plugging items into power strips
- Filling a lunchbox
- Moving plastic vegetables
- Zipping up bags
These actions were performed in response to verbal commands, illustrating the interactive and adaptive qualities of the new robotic systems.
Key Features of Gemini Robotics
According to Google, AI models for robotics should possess three essential qualities to be effective:
- General: These models must be able to adjust to various situations and environments.
- Interactive: The AI should quickly comprehend and respond to instructions or changes around them.
- Dexterous: Robots need to manipulate objects in a way similar to how humans do with their hands.
These attributes are crucial for making robots genuinely helpful in everyday scenarios.
Focus on Customization with Gemini Robotics-ER
The Gemini Robotics-ER model serves as a foundation for robotic engineers, allowing them to create customized AI solutions tailored to their specific needs. This version is accessible not only to Apptronik but also to a select group of "trusted testers," including companies like Boston Dynamics and Agile Robots. This approach encourages innovation and collaboration within the robotics community.
The Competitive Landscape in Robotics and AI
Google is not the only player in the race to integrate AI into robotics. In November, OpenAI invested in a startup called Physical Intelligence, which focuses on deploying general-purpose AI in physical settings. This startup aims to develop large-scale AI models for robotic applications, highlighting the growing intersection between AI and physical tasks.
In a similar vein, OpenAI has also recruited experts from organizations like Meta to spearhead its robotics and consumer hardware efforts. Tesla is also making strides in this domain with its Optimus robot, reflecting the rising interest and investment in humanoid robotics.
Google’s Vision for Robotics
Sundar Pichai, the CEO of Google, expressed that the company views robotics as an excellent opportunity for applying and testing AI technologies in the physical realm. He emphasized that the new robots would leverage Google’s multimodal AI capabilities to adapt in real-time to their surroundings, showcasing a promising future for robotics enhanced by artificial intelligence technologies.
The development of these robotics models by Google represents a significant step towards creating smart, capable robots that could eventually play essential roles in various industries, enhancing efficiency and automation. As technology continues to evolve, the potential applications of AI-powered robots are bound to expand, opening new avenues for both innovation and collaboration in this exciting field.
