The dataset was proposed by Google researchers in 2022, and the Robotics Transformer model (RT-1) was launched at the same time, a multi-task model that has achieved significant improvements in zero-shot generalization on new tasks, environments, and objects, demonstrating excellent scalability and pre-trained model characteristics.

RT-1 RobotAction Dataset A large-scale real-world robotics dataset for training the RT-1 model. The researchers used 13 EDR robotic arms, each equipped with a 7-DOF arm, a two-finger gripper, and a mobile base, and collected 130,000 clips in 17 months, totaling 111.06 GB, each of which was annotated with a text description of the robot's execution instructions. The high-level skills covered in the dataset include picking up and placing objects, opening and closing drawers, taking objects out of and putting them in drawers, placing slender objects vertically, pushing down objects, pulling napkins, and opening cans, covering more than 700 tasks using a variety of different objects.