arx5
Datasets
All datasets matching “arx5”arx5-robot-dataset
ARX5 Manipulation Dataset
Robotic manipulation demos collected on a single ARX5 arm, of tabletop tasks with dual RGB cameras and DINOv2 embeddings of the images.
Converted from Ville Kuosmanen's LeRobot v2.1 datasets into a unified HDF5 format.
Hardware
Robot: ARX5 single arm (7-DoF)
Cameras: Wrist-mounted + front-facing scene camera
Stats
Trajectories
1,704
Total frames
860,076
Frames per trajectory
4 – 3,034 (median 397)
Total size
1 TB+… See the full description on the dataset page: https://huggingface.co/datasets/pravsels/arx5-robot-dataset.evButtonPush-arx5-260715-0-MCAPevButtonPush-arx5-260722-0-lerobot
evButtonPush arx5 v3 - LeRobot dataset
ARX X5 6-DoF arm pressing an elevator button, leader-follower teleoperation.
Used to fine-tune GR00T N1.6.
Episodes: 112 (odd/even split of 56 recordings, 30fps -> two real 15fps sequences)
Frames: 22,518 | FPS: 15 | State/Action: 7D (6 joints + gripper) | Cameras: 3
Paired inference rosbag: evButtonPush-arx5-260722-0-MCAP
robochallenge_dump_arx5This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "arx5",
"total_episodes": 1,
"total_frames": 3740,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 30,
"splits": {
"train": "0:1"
},
"data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/typoverflow/robochallenge_dump_arx5.arx5_real_world_datasets
arx5_real_world_datasets
This repository contains the real-world robot datasets collected on the ARX-5 platform as presented in the paper Self-Correcting VLA: Online Action Refinement via Sparse World Imagination.
The dataset is part of the SC-VLA framework, which enhances robotic physical grounding through intrinsic self-improvement using Sparse World Imagination (SPI) and Online Action Refinement (OAR).
Code: https://github.com/Kisaragi0/SC-VLA
Paper: Self-Correcting VLA: Online… See the full description on the dataset page: https://huggingface.co/datasets/Kisaragi0/arx5_real_world_datasets.robotwin-arx5-v4mix-train
RoboTwin ARX-X5 pi0.5 v4mix training dataset
Exact training mixture used by pi05_robotwin_icl_arx_x5_v4mix:
3,000 canonical RoboTwin ICL-training episodes across 20 tasks.
1,500 scripted-variant episodes across 10 variant cells.
4,500 episodes total; head, left-wrist, and right-wrist cameras at 25 FPS.
LeRobot dataset format v2.1.
The five canonical ICL holdout tasks and all source validation episodes are excluded.
Original source provenance is preserved in… See the full description on the dataset page: https://huggingface.co/datasets/yeeeiii111/robotwin-arx5-v4mix-train.