CoolFace
Datasetpublic

una-auxme/ROS2SmolVLA_ur10e_no_joints_crop_pick_place

This is the training dataset for our ROS2SmolVLA project It was recorded by teleoperation of our UR10e lightweight industrial robot through our ROS2SmolVLA setup utilizing LeRobot. The action space is: "linear_x.vel", "linear_y.vel", "linear_z.vel", "angular_x.vel", "angular_y.vel", "angular_z.vel", "gripper.pos" The observation space is: "pose.x", "pose.y", "pose.z", "pose.quat_x", "pose.quat_y", "pose.quat_z", "pose.quat_w" one 720x720 and two 1280x720 camera streams.… See the full description on the dataset page: https://huggingface.co/datasets/una-auxme/ROS2SmolVLA_ur10e_no_joints_crop_pick_place.

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face
1likes286downloads
Dataset Card

This is the training dataset for our ROS2SmolVLA project

It was recorded by teleoperation of our UR10e lightweight industrial robot through our ROS2SmolVLA setup utilizing LeRobot.

The action space is:

"linear_x.vel",
"linear_y.vel",
"linear_z.vel",
"angular_x.vel",
"angular_y.vel",
"angular_z.vel",
"gripper.pos"

The observation space is:

"pose.x",
"pose.y",
"pose.z",
"pose.quat_x",
"pose.quat_y",
"pose.quat_z",
"pose.quat_w"

one 720x720 and two 1280x720 camera streams.

<a class="flex" href="https://huggingface.co/spaces/lerobot/visualizedataset?path=una-auxme/ROS2SmolVLAur10enojointscroppick_place"> <img class="block dark:hidden" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/visualize-this-dataset-xl.svg"/> <img class="hidden dark:block" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/visualize-this-dataset-xl-dark.svg"/> </a>

Dataset Description

  • —Homepage: ROS2SmolVLA
  • —Paper: [More Information Needed]
  • —License: apache-2.0

Dataset Structure

meta/info.json:

json
{
    "codebase_version": "v3.0",
    "fps": 30,
    "features": {
        "action": {
            "dtype": "float32",
            "names": [
                "linear_x.vel",
                "linear_y.vel",
                "linear_z.vel",
                "angular_x.vel",
                "angular_y.vel",
                "angular_z.vel",
                "gripper.pos"
            ],
            "shape": [
                7
            ]
        },
        "observation.images.camera1": {
            "dtype": "video",
            "shape": [
                720,
                720,
                3
            ],
            "names": [
                "height",
                "width",
                "channels"
            ],
            "info": {
                "video.height": 1280,
                "video.width": 720,
                "video.codec": "av1",
                "video.pix_fmt": "yuv420p",
                "video.is_depth_map": false,
                "video.fps": 30,
                "video.channels": 3,
                "has_audio": false,
                "video.video_backend": "pyav",
                "video.preset": null,
                "video.extra_options": null,
                "video.crf": null,
                "video.fast_decode": null,
                "video.g": null,
                "video_height": 720
            }
        },
        "observation.images.camera2": {
            "dtype": "video",
            "shape": [
                720,
                1280,
                3
            ],
            "names": [
                "height",
                "width",
                "channels"
            ],
            "info": {
                "video.height": 720,
                "video.width": 1280,
                "video.codec": "av1",
                "video.pix_fmt": "yuv420p",
                "video.is_depth_map": false,
                "video.fps": 30,
                "video.channels": 3,
                "has_audio": false,
                "video.video_backend": "pyav",
                "video.preset": null,
                "video.extra_options": null,
                "video.crf": null,
                "video.fast_decode": null,
                "video.g": null
            }
        },
        "observation.images.camera3": {
            "dtype": "video",
            "shape": [
                720,
                1280,
                3
            ],
            "names": [
                "height",
                "width",
                "channels"
            ],
            "info": {
                "video.height": 720,
                "video.width": 1280,
                "video.codec": "av1",
                "video.pix_fmt": "yuv420p",
                "video.is_depth_map": false,
                "video.fps": 30,
                "video.channels": 3,
                "has_audio": false,
                "video.video_backend": "pyav",
                "video.preset": null,
                "video.extra_options": null,
                "video.crf": null,
                "video.fast_decode": null,
                "video.g": null
            }
        },
        "timestamp": {
            "dtype": "float32",
            "shape": [
                1
            ],
            "names": null
        },
        "frame_index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "episode_index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "task_index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "observation.state": {
            "dtype": "float32",
            "names": [
                "pose.x",
                "pose.y",
                "pose.z",
                "pose.quat_x",
                "pose.quat_y",
                "pose.quat_z",
                "pose.quat_w"
            ],
            "shape": [
                7
            ]
        }
    },
    "total_episodes": 349,
    "total_frames": 325838,
    "total_tasks": 4,
    "chunks_size": 1000,
    "data_files_size_in_mb": 100,
    "video_files_size_in_mb": 200,
    "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
    "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4",
    "robot_type": null,
    "splits": {
        "train": "0:349"
    }
}

Citation

BibTeX:

bibtex
@misc{mandischer2026ros2smolvlaenablingsmallvisionlanguageaction,
      title={ROS2SmolVLA: Enabling Small Vision-Language-Action Models for Integration into Industrial-Grade Lightweight Robots}, 
      author={Nils Mandischer and Noah Böckmann and Ludwig Holl and Lars Mikelsons},
      year={2026},
      eprint={2608.23320},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2608.23320}, 
}