In a groundbreaking announcement, Google DeepMind has introduced two new AI models designed to revolutionize the field of robotics. These models, Gemini Robotics and Gemini Robotics-ER, are set to enhance the capabilities of robots, enabling them to interact with objects, navigate environments, and perform complex tasks with unprecedented precision.
Gemini Robotics: A Leap in Vision-Language-Action Systems
Gemini Robotics, the first of the two models, is a vision-language-action (VLA) system that can adapt to new situations without prior training. Built on the foundation of Gemini 2.0, this model integrates physical actions as a new modality, allowing robots to perform tasks such as folding paper or opening a bottle cap with improved dexterity and precision.
Gemini Robotics-ER: Advanced Spatial Understanding
The second model, Gemini Robotics-ER, offers advanced spatial understanding, enabling robots to comprehend complex and dynamic environments. This model is designed to work with existing low-level controllers, providing roboticists with the tools to develop new capabilities for their machines.
Enhancing Human-Robot Interaction
One of the key features of these new models is their ability to interact seamlessly with humans and their surroundings. Gemini Robotics can understand and respond to everyday commands, adapting its behaviour to user input and changes in the environment. This interactivity makes it possible for robots to perform a wide range of real-world tasks, from packing a lunchbox to preparing a salad.
A Focus on Safety and Collaboration
Google DeepMind is committed to ensuring the safety of its AI models. The company has introduced new benchmarks and frameworks to enhance safety research within the AI industry. Additionally, DeepMind is partnering with companies like Apptronik to build the next generation of humanoid robots, further expanding the potential applications of these advanced AI models.
Conclusion
The unveiling of Gemini Robotics and Gemini Robotics-ER marks a significant milestone in the field of robotics. These models bring AI into the physical world, enabling robots to perform tasks with greater generality, interactivity, and dexterity. As Google DeepMind continues to push the boundaries of AI, the future of robotics looks more promising than ever.