mpark's picture
Upload README.md with huggingface_hub
021403a verified
|
Raw
History Blame Contribute Delete
6.8 kB
metadata
license: apache-2.0
task_categories:
  - robotics
tags:
  - LeRobot
  - robotics
  - lerobot
  - manipulation
  - franka
  - isaac-lab
  - simulation
  - oracle
configs:
  - config_name: default
    data_files: data/*/*.parquet

This dataset was created using LeRobot.

Dataset Description

Simulated manipulation demonstrations generated in RoboLab (Isaac Lab) by a deterministic waypoint oracle with privileged state access (no cameras used for control). Task: FoodPacking1Cans, difficulty level medium.

What this is

Part of a 3 task x 3 difficulty grid of oracle datasets. The oracle replays a fixed waypoint recipe solved with rest-cost IK (pyroki) and is rate-limited to match the arm-speed cap of the corresponding RL policy, so the action statistics stay in the same regime as learned policies.

Difficulty

Initial object poses are drawn from a pre-generated layout dataset. The level controls two axes:

level position fraction yaw range
soft 0.20 x feasible 0 deg
medium 0.50 x feasible +/- 25 deg
hard 1.00 x feasible +/- 180 deg

Oracle success rate on this level: 0.61 (measured, n>=128 attempts). Only successful episodes are stored.

Format

  • LeRobot v3.0 (tag v3.0), 15 fps
  • Two cameras at 270x480: observation.images.wrist_cam, observation.images.over_shoulder_left_camera
  • Videos are H.264 / yuv420p and CFR-normalized (constant PTS delta). Non-CFR video breaks torchcodec-based training loaders.
  • q01/q99 quantile statistics are computed once after merging all shards.

Action convention (read this before training)

Actions are absolute DROID-frame joint targets (7 arm joints, j7 offset -pi/4), not deltas. The gripper channel is binary {0, 1}, not continuous. A continuous gripper signal is known to overestimate downstream policy performance on this benchmark, which is why it is binarized at collection time.

Success requires the object to remain in the container for a 15-step hold, so the final frames of each episode are a settle phase rather than the moment of release.

Provenance

meta/provenance.json records the oracle identity, the recipe reference, the rate-limit vector, and the initial-pose level. Episodes were collected without cameras and rendered afterwards by teleporting through the saved states, so the same trajectories can be re-rendered at other resolutions or camera poses.

License

Apache-2.0. Generated entirely in simulation; contains no third-party recorded data.

Visualize

Open in the LeRobot dataset visualizer

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

Dataset Structure

meta/info.json:

{
    "codebase_version": "v3.0",
    "robot_type": "unknown",
    "total_episodes": 3095,
    "total_frames": 467310,
    "total_tasks": 1,
    "chunks_size": 1000,
    "data_files_size_in_mb": 100,
    "video_files_size_in_mb": 200,
    "fps": 15,
    "splits": {
        "train": "0:3095"
    },
    "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",
    "features": {
        "action": {
            "dtype": "float32",
            "shape": [
                8
            ],
            "names": {
                "motors": [
                    "j0",
                    "j1",
                    "j2",
                    "j3",
                    "j4",
                    "j5",
                    "j6",
                    "gripper"
                ]
            },
            "fps": 15
        },
        "observation.state": {
            "dtype": "float32",
            "shape": [
                8
            ],
            "names": {
                "motors": [
                    "joint_0",
                    "joint_1",
                    "joint_2",
                    "joint_3",
                    "joint_4",
                    "joint_5",
                    "joint_6",
                    "joint_7"
                ]
            },
            "fps": 15
        },
        "observation.velocity": {
            "dtype": "float32",
            "shape": [
                8
            ],
            "names": null,
            "fps": 15
        },
        "episode_index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "frame_index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "task_index": {
            "dtype": "int64",
            "shape": [
                1
            ],
            "names": null
        },
        "timestamp": {
            "dtype": "float32",
            "shape": [
                1
            ],
            "names": null
        },
        "next.done": {
            "dtype": "bool",
            "shape": [
                1
            ],
            "names": null,
            "fps": 15
        },
        "observation.images.over_shoulder_left_camera": {
            "dtype": "video",
            "shape": [
                270,
                480,
                3
            ],
            "names": [
                "height",
                "width",
                "channels"
            ],
            "info": {
                "video.height": 270,
                "video.width": 480,
                "video.codec": "avc1",
                "video.pix_fmt": "yuv420p",
                "video.is_depth_map": false,
                "video.fps": 15.0,
                "video.channels": 3,
                "has_audio": false
            }
        },
        "observation.images.wrist_cam": {
            "dtype": "video",
            "shape": [
                270,
                480,
                3
            ],
            "names": [
                "height",
                "width",
                "channels"
            ],
            "info": {
                "video.height": 270,
                "video.width": 480,
                "video.codec": "avc1",
                "video.pix_fmt": "yuv420p",
                "video.is_depth_map": false,
                "video.fps": 15.0,
                "video.channels": 3,
                "has_audio": false
            }
        }
    }
}

Citation

BibTeX:

[More Information Needed]