Datasets:
metadata
license: apache-2.0
task_categories:
- robotics
tags:
- LeRobot
- robotics
- lerobot
- vla
configs:
- config_name: default
data_files: data/*/*.parquet
This dataset was created using LeRobot.
Dataset Description
- Homepage: [More Information Needed]
- Paper: [More Information Needed]
- License: apache-2.0
Dataset Structure
{
"codebase_version": "v3.0",
"robot_type": "seeed_b601_rs_follower",
"total_episodes": 100,
"total_frames": 42746,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 30,
"splits": {
"train": "0:100"
},
"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",
"names": [
"shoulder_pan.pos",
"shoulder_lift.pos",
"elbow_flex.pos",
"wrist_flex.pos",
"wrist_yaw.pos",
"wrist_roll.pos",
"gripper.pos"
],
"shape": [
7
]
},
"observation.state": {
"dtype": "float32",
"names": [
"shoulder_pan.pos",
"shoulder_lift.pos",
"elbow_flex.pos",
"wrist_flex.pos",
"wrist_yaw.pos",
"wrist_roll.pos",
"gripper.pos"
],
"shape": [
7
]
},
"observation.images.wrist": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height": 480,
"video.width": 640,
"video.codec": "h264",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"video.fps": 30,
"video.channels": 3,
"has_audio": false
}
},
"observation.images.wrist_depth": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height": 480,
"video.width": 640,
"video.codec": "h264",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"video.fps": 30,
"video.channels": 3,
"has_audio": false
}
},
"observation.images.overhead": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height": 480,
"video.width": 640,
"video.codec": "h264",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"video.fps": 30,
"video.channels": 3,
"has_audio": false
}
},
"observation.images.overhead_depth": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height": 480,
"video.width": 640,
"video.codec": "h264",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"video.fps": 30,
"video.channels": 3,
"has_audio": false
}
},
"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:
[More Information Needed]
Depth (RGB-D)
This dataset includes raw metric depth alongside the RGB videos above, for overhead, wrist. It is not part of the standard LeRobot schema, so generic loaders will not see it automatically.
depth/<camera>_depth/episode_NNNNNN.mkv FFV1, gray16le, lossless uint16
meta/depth_info.json depth_scale_m + intrinsics per camera
Depth is aligned to its colour camera on capture, so depth[y, x] and the
matching observation.images.<camera>[y, x] are the same physical point.
metres = pixel_value * depth_scale_m; 0 means no return.
Read it with read_depth.py from the collection tooling used to record this
dataset, or directly:
import av, json, numpy as np
info = json.load(open("meta/depth_info.json"))
with av.open("depth/wrist_depth/episode_000000.mkv") as c:
depth = np.stack([f.to_ndarray() for f in c.decode(video=0)])
depth_m = depth.astype(np.float32) * info["cameras"]["wrist_depth"]["depth_scale_m"]