DeepMind Announces Plans to Utilize AI Models for Operating Physical Robots
Google DeepMind’s Advances in Robotics
Google DeepMind recently introduced two advanced AI models designed specifically for robotics, highlighting a significant leap in integrating artificial intelligence with physical machines. This development is based on the Gemini 2.0 platform, which Google claims is its most sophisticated AI yet.
New AI Models for Robotics
The two new models unveiled are known as Gemini Robotics and Gemini Robotics-ER (Extended Reasoning). Both are powered by Gemini 2.0, which extends beyond AI’s traditional focus on generating text and images to executing physical commands aimed at controlling robots. This shift marks an important step towards enabling robots to perform complex tasks in real-world environments.
Collaboration with Apptronik
Google is partnering with Apptronik, a robotics developer based in Texas, to create a new generation of humanoid robots. Apptronik has a history of collaboration with notable organizations like Nvidia and NASA. Recently, Google participated in a $350 million funding round for Apptronik, showcasing its commitment to advancing robotics technology.
Demonstrations of Capabilities
In recent demonstration videos, the robots equipped with Gemini’s AI showcased an array of tasks. These include plugging devices into power strips, organizing items in a lunchbox, and moving plastic vegetables. Each action was prompted by spoken commands, illustrating the robots’ ability to execute complex tasks in a dynamic environment.
Essential Qualities of Robotics AI
According to Google, effective AI models for robotics must exhibit three core qualities:
- Generalization: The ability to adapt to various situations.
- Interactivity: Capability to comprehend and respond promptly to instructions or changes in their environment.
- Dexterity: The skill to manipulate objects in ways similar to human hands.
This framework aims to ensure that AI models can operate effectively while engaging in tasks that require nuanced physical actions.
Gemini Robotics-ER for Developers
The Gemini Robotics-ER model is tailored for roboticists to build upon and develop their own AI solutions. Open to Apptronik and select trusted testers such as Agile Robots, Agility Robots, Boston Dynamics, and Enchanted Tools, this version serves as a foundational tool for companies looking to enhance their robotic capabilities.
The Competitive Landscape in AI and Robotics
Google is not alone in its pursuit of integrating AI into robotics. Other companies are also making strides in this area. For example, OpenAI recently invested in Physical Intelligence, a startup focused on creating general-purpose AI for robotics. Additionally, OpenAI has been strengthening its engineering team by hiring industry veterans to spearhead its robotics initiatives.
Tesla has also entered the humanoid robotics domain with its Optimus robot, demonstrating the growing interest and investment in robotic technologies across various sectors.
The Future of Robotics at Google
Google CEO Sundar Pichai has expressed the company’s vision of utilizing robotics as a proving ground for AI advancements that can be applied in physical contexts. The robots will leverage Google’s multimodal AI capabilities to adapt to their surroundings in real-time, indicating the potential for more context-aware robotic systems in the future.
This focus on robotics not only showcases Google DeepMind’s technological prowess but also highlights the broader trend of enhancing AI’s role in real-world applications. The ongoing developments are set to redefine the landscape of robotics, combining intelligence with functionality to meet the needs of various industries.
