Google DeepMind is launching two new AI models that will help robots “perform a wider range of real-world tasks than ever before.” The first, called Gemini Robotics, is a vision-speech-action model that can understand new situations even if it hasn’t been trained on them.
Google DeepMind is launching two new AI models that will help robots “perform a wider range of real-world tasks than ever before.” The first, called Gemini Robotics, is a vision-speech-action model that can understand new situations even if it hasn’t been trained on them.
Gemini Robotics is built on Gemini 2.0, The Verge reports . This model has made progress in three key areas that Google DeepMind believes are essential for creating useful robots: generalization, interactivity, and agility.
In addition to being able to generalize to new scenarios, Gemini Robotics is better able to interact with people and the environment. It is also able to perform more precise physical tasks, such as folding a sheet of paper or removing a cap from a bottle.
“This allows us to create robots that are more capable, more responsive, and more resilient to changes in their environment,” says Carolina Parada, senior director and head of robotics at Google DeepMind.
According to Google DeepMind, Gemini Robotics-ER is an advanced visual language model capable of “understanding our complex and dynamic world.”
As Parada explains, when you're packing a lunch box and there are items on the table in front of you, you need to know where everything is, how to open the lunch box, how to get the items, and where to put them. These are the considerations that Gemini Robotics-ER should be able to handle.
This model is designed for roboticists to connect to existing low-level controllers—the system that controls the robot's movements—allowing them to take advantage of the new capabilities provided by Gemini Robotics-ER.
