CoolFace
Datasetpublic

Nfiniteai/so_arm_100_image_rnd_2

This dataset was created using LeRobot. Dataset Description The SO ARM 100 test dataset series is an example dataset for testing purposes. It is a synthetic dataset from a high-fidelity pick and place task recorded from an expert policy trained in IsaacLab. The dataset consists of image observations, joint angle observations, and joint angle actions. There are five datasets in the series with varying degrees of domain randomization, as shown below. In addition to the state… See the full description on the dataset page: https://huggingface.co/datasets/Nfiniteai/so_arm_100_image_rnd_2.

sourceHugging Faceapache-2.0updated 1y agoView on Hugging Face
0likes70downloads
Dataset Card

This dataset was created using LeRobot.

Dataset Description

The SO ARM 100 test dataset series is an example dataset for testing purposes. It is a synthetic dataset from a high-fidelity pick and place task recorded from an expert policy trained in IsaacLab. The dataset consists of image observations, joint angle observations, and joint angle actions. There are five datasets in the series with varying degrees of domain randomization, as shown below. In addition to the state randomization, the dataset contains all the previous randomizations, starting with so_arm_100_image_rnd_2.

Dataset NameRandomization
so_arm_100_image_no_rndNo randomizations
so_arm_100_image_rnd_1Cube table pose
so_arm_100_image_rnd_2+ Plate table position
so_arm_100_image_rnd_3+ Camera pose
so_arm_100_image_rnd_4+ Light intensity

Dataset Structure

meta/info.json:

json
{
    "codebase_version": "v2.1",
    "robot_type": "simulated_so_arm_100",
    "total_episodes": 200,
    "total_frames": 31208,
    "total_tasks": 1,
    "total_videos": 0,
    "total_chunks": 1,
    "chunks_size": 1000,
    "fps": 30.0,
    "splits": {
        "train": "0:200"
    },
    "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
    "video_path": "videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4",
    "features": {
        "observation.image": {
            "dtype": "image",
            "names": [
                "channels",
                "height",
                "width"
            ],
            "shape": [
                3,
                512,
                512
            ]
        },
        "observation.state": {
            "dtype": "float32",
            "shape": [
                6
            ]
        },
        "action": {
            "dtype": "float32",
            "shape": [
                6
            ]
        },
        "next.success": {
            "dtype": "bool",
            "shape": [
                1
            ]
        },
        "next.reward": {
            "dtype": "float32",
            "shape": [
                1
            ]
        },
        "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
        }
    }
}

Citation

BibTeX:

bibtex
@inproceedings{nfinite-ds-1,
  author = {Miguel Alonso Jr},
  editor = {nfinite R&D Team},
  title = {nfinite physical AI: SO ARM 100 Test Data},
  year = {2025}
}