Google’s DeepMind team recently debuted Open X-Embodiment, a database of robotics functionality created in collaboration with 33 research institutes. The database contains 500+ skills and 150,000 tasks gathered from 22 robot embodiments. DeepMind then trained its RT-1-X model on the data and used it to train robots in other labs, reporting a 50% success rate compared to the in-house methods the teams had developed. It is an exciting time for robotic learning, with many teams approaching the problem from different angles with ever-increasing efficacy. Simulation and AI, including the generative variety, will be a big part of the equation for general-purpose robots.
