On Friday, Google DeepMind announced Robotic Transformer 2 (RT-2), a “first-of-its-kind” vision-language-action (VLA) model that uses data scraped from the Internet to enable better robotic control ...