RT-2: New model translates vision and language into action

ROBOCAT is a self-improving robotic agent designed to perform specific tasks well, but it is limited by its current programming. The researchers explore the possibility of stacking pebbles and logs on a stone using their robot, a process that seems simple for a person but requires careful handling due to the differences in their behavior. They acknowledge that their information will be used in accordance with Google’s Privacy Policy. Keywords: Robots, AI, robotics, generalization, machine learning, computer vision, programming, stacking pebbles and logs on a stone