Google’s AI research lab, DeepMind, has announced a new AI model called Gemini Robotics designed to enable real-world machines to interact with objects and environments. The model allows robots to perform tasks in response to voice commands, such as folding paper and putting on glasses.
Gemini Robotics was trained to generalize behavior across different robotics hardware and connect visual inputs with actions. In tests, the model performed well even outside of its training data range. To facilitate further research, DeepMind has released a slimmed-down version of Gemini Robotics-ER and a benchmark called Asimov to assess risks associated with AI-powered robots.
The release of Gemini Robotics marks an important step forward for robotics control using artificial intelligence, offering a promising solution for future advancements in the field.
Source: https://techcrunch.com/2025/03/12/google-deepmind-unveils-new-ai-models-for-controlling-robots