An RGBD robotic manipulation dataset regenerated from RoboTwin 2.0, augmented with depth camera data for training imitation learning policies such as PI-0. Uses the LargeD435 camera configuration (640×480 resolution), matching the raw shape in configs/data/robotwin.yaml. Both versions cover all 50 manipulation tasks in RoboTwin 2.0, with depth: true and LargeD435 cameras (640×480). Each camera provides both RGB and depth (uint16, millimeters) data. Raw collected data is stored as HDF5 files. Each episode contains: - jointaction/ — left/right arm joint positions + gripper state - observation/ /rgb — RGB images (JPEG encoded) - observation/ /depth — depth maps (float32, millimeters)…