Google DeepMind’s new AI model can control a robot’s entire body

Google DeepMind has announced the release of its latest AI model, Gemini Robotics 2, which is capable of controlling entire humanoid robots. This new version marks a significant advancement from its predecessor, which was limited to controlling only the upper body of humanoid robots. Gemini Robotics 2 now enables "whole-body motions," allowing robots to perform a broader range of actions, including walking, crouching, stretching, and manipulating objects. Demonstrations of the capabilities of Gemini Robotics 2 feature Apptronik's Apollo 2 robot. Videos shared by Google illustrate the robot's ability to bend over and pick up a watering can, as well as locate and retrieve specific items from a shelf. This enhanced functionality signifies a leap towards what is referred to as "physical AGI" (Artificial General Intelligence), indicating a more sophisticated integration of AI into physical tasks. While the advancements in Gemini Robotics 2 present exciting possibilities for the use of AI in real-world applications, there are also associated risks. The introduction of AI into physical environments raises concerns that need to be addressed as the technology develops. The implications of these advancements in robotics and AI are significant, as they could transform various industries and everyday tasks.