123qwe55aa/my_rover_encoder_timing_20261011

This dataset was created using LeRobot.

Dataset Description

Public personal research dataset from the lerobot-lab rover project.

• Source: physical LeKiwi-style three-omni-wheel rover.
• Robot type: real_rover.
• Episodes: 1; frames: 1063; nominal FPS: 30.
• Format: LeRobot v3.0; created/checked using installed LeRobot 0.6.2.
• Image feature: observation.images.front; see meta/info.json for dimensions and encoding.
• Action: [x.vel, y.vel, theta.vel], in [m/s, m/s, degrees/s] (body frame: forward, left, counterclockwise yaw).

Meaning of observation.state

observation.state contains the previous sent body-velocity command, in the same order and units as action. It is not measured wheel encoder velocity and is not body velocity reconstructed from encoders. Observation is collected before sending the current action. Do not describe this feature as measured proprioception.

Dataset purpose

Independent timing diagnosis. Keep separate from imitation-learning training demonstrations.

Real collection snapshots are preserved separately; later merged versions may overlap their source episodes. Do not count all versions as independent demonstrations.

License and visibility

Publicly accessible personal research archive. No redistribution license is granted by this upload. The LeRobot software license does not automatically license these recordings.

Payload verification

meta/, data/ and videos/ were uploaded intact and verified against local SHA256 hashes. Local validation checks episode metadata/frame totals and first/last sample loading; it is not a task-success certification.

Timing sidecar

meta/timing/session-*.jsonl records host monotonic timing and saved/discarded episode attempts. Timestamps are local time.perf_counter values, not camera exposure timestamps. Only episode_saved confirms which attempt was committed; exclude discarded attempts.

Encoder diagnostic readings are raw Present_Position counts for three wheel motors inside timing logs. Speed was estimated offline by differentiating unwrapped position; these diagnostic values do not replace observation.state. Host timestamps/encoder feedback are not a calibrated ground-truth chassis pose or speed sensor.

• Homepage: [More Information Needed]
• Paper: [More Information Needed]
• License: [More Information Needed]

Dataset Structure

meta/info.json:

{
    "codebase_version": "v3.0",
    "fps": 30,
    "features": {
        "observation.state": {
            "dtype": "float32",
            "shape": [
                3
            ],
            "names": [
                "x.vel",
                "y.vel",
                "theta.vel"
            ]
        },
        "observation.images.front": {
            "dtype": "video",
            "shape": [
                480,
                640,
                3
            ],
            "names": [
                "height",
                "width",
                "channels"
            ],
            "info": {
                "is_depth_map": false,
                "video.height": 480,
                "video.width": 640,
                "video.codec": "h264",
                "video.pix_fmt": "yuv420p",
                "video.fps": 30,
                "video.channels": 3,
                "has_audio": false,
                "video.g": 2,
                "video.crf": 30,
                "video.preset": "ultrafast",
                "video.fast_decode": 0,
                "video.video_backend": "pyav",
                "video.extra_options": {}
            }
        },
        "action": {
            "dtype": "float32",
            "shape": [
                3
            ],
            "names": [
                "x.vel",
                "y.vel",
                "theta.vel"
            ]
        },
        "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
        }
    },
    "total_episodes": 1,
    "total_frames": 1063,
    "total_tasks": 1,
    "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": "real_rover",
    "splits": {
        "train": "0:1"
    }
}

Citation

BibTeX:

[More Information Needed]