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

Google DeepMind has released an updated version of its Gemini Robotics AI model that can control a humanoid robot's entire body, including movements from the feet to the fingertips, whereas the previous version only controlled the upper body. The new model enables robots to perform a wider range of physical tasks such as walking, crouching, stretching, picking up objects, and manipulating items with five-fingered hands, allowing them to complete more complex real-world tasks that require whole-body coordination. An updated companion model called Gemini Robotics ER 2 helps robots analyze their surroundings and complete multi-step tasks over extended periods, and includes safety improvements to detect nearby humans and stop the robot if someone approaches too closely. The company also improved its On-Device Model, which can run locally on a robot without requiring an internet connection.