Google has introduced two advanced variants of its Gemini AI system, expanding its impact from the digital to the physical world by enabling robots to be controlled on demand. The first variant, Gemini Robotics, enhances the AI’s capability to produce content while managing robotic actions. The second, dubbed “ER” (embodied reasoning), boosts Gemini’s spatial reasoning skills.
Gemini Robotics-ER can learn new skills quickly, allowing it to manipulate objects like a coffee cup with precision. Collaborating with Texan robotics start-up Apptronik, Google aims to develop efficient humanoid robots capable of performing tasks in dynamic settings, exemplified by the humanoid robot “Apollo.”
Essential properties of these AI models include versatility, interactivity, and dexterity, drastically improving performance in robotics. Google plans to test their advanced AI with various robotics specialists, while prioritizing safety measures to ensure secure interactions between robots and humans.
Source link
