Cache-SCA/smolVLA_UR7e_PickandPlace_50epoch_tp1
Robotics • 0.5B • Updated • 37
Datasets:
100 episodes · 10 fps · 2 cameras · 640×480 h264
This dataset was created using LeRobot.
Dataset Description
10fps downsampled LeRobot v3 dataset for UR7e red-block-on-blue-dish pick-and-place. It is derived from CoRL2026-CSI/UR7e_CaP_PickandPlace_100epi by keeping source frames 0, 3, 6, ... in each successful forward episode. Numeric observations and absolute 7D UR7e actions are copied from the retained source frames without interpolation.
- Homepage: https://huggingface.co/datasets/CoRL2026-CSI/UR7e_CaP_PickandPlace_100epi
- Paper: [More Information Needed]
- License: apache-2.0
Dataset Structure
{
"codebase_version": "v3.0",
"robot_type": "ur7e",
"total_episodes": 100,
"total_frames": 35878,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"fps": 10,
"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": {
"observation.state": {
"dtype": "float32",
"shape": [
7
],
"names": [
"shoulder_pan.pos",
"shoulder_lift.pos",
"elbow.pos",
"wrist_1.pos",
"wrist_2.pos",
"wrist_3.pos",
"gripper.pos"
]
},
"action": {
"dtype": "float32",
"shape": [
7
],
"names": [
"shoulder_pan.pos",
"shoulder_lift.pos",
"elbow.pos",
"wrist_1.pos",
"wrist_2.pos",
"wrist_3.pos",
"gripper.pos"
]
},
"observation.images.realsense_topview": {
"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": 10,
"video.channels": 3,
"has_audio": false
}
},
"observation.images.realsense_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": 10,
"video.channels": 3,
"has_audio": false
}
},
"skill.type": {
"dtype": "string",
"shape": [
1
],
"names": null
},
"skill.progress": {
"dtype": "float32",
"shape": [
1
],
"names": null
},
"skill.goal_position.joint": {
"dtype": "float32",
"shape": [
7
],
"names": [
"shoulder_pan.pos",
"shoulder_lift.pos",
"elbow.pos",
"wrist_1.pos",
"wrist_2.pos",
"wrist_3.pos",
"gripper.pos"
]
},
"skill.goal_position.world_xyzrpy": {
"dtype": "float32",
"shape": [
6
],
"names": [
"x",
"y",
"z",
"roll",
"pitch",
"yaw"
]
},
"skill.goal_position.robot_xyzrpy": {
"dtype": "float32",
"shape": [
6
],
"names": [
"x",
"y",
"z",
"roll",
"pitch",
"yaw"
]
},
"skill.goal_position.gripper": {
"dtype": "float32",
"shape": [
1
],
"names": [
"gripper.pos"
]
},
"skill.natural_language": {
"dtype": "string",
"shape": [
1
],
"names": null
},
"observation.ee_pos.robot_xyzrpy": {
"dtype": "float32",
"shape": [
6
],
"names": [
"x",
"y",
"z",
"roll",
"pitch",
"yaw"
]
},
"observation.gripper_binary": {
"dtype": "float32",
"shape": [
1
],
"names": 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
}
}
}
Citation
BibTeX:
[More Information Needed]
Action Relabeling
This derivative dataset rewrites the action column from future observed robot state.
- Source:
CoRL2026-CSI/UR7e_CaP_PickandPlace_100epi_10fps - Target repo:
CoRL2026-CSI/UR7e_CaP_PickandPlace_100epi_10fps_state_tplus1_action - Formula:
action[t] = observation.state[t + 1]within each episode - Terminal handling: final episode frames are clamped to the final observed state
- Original videos, tasks, timestamps, and feature schema are preserved
- Downloads last month
- 84
Models trained or fine-tuned on Cache-SCA/UR7e_CaP_PickandPlace_100epi_10fps_state_tplus1_action