DaivdYuan/hub-strap-wires-with-velcro-tape-lerobot
收藏资源简介：
--- license: other tags: - robotics - lerobot - manipulation - unknown - nmotion-initiative --- # Strap wires with velcro tape ## Visualizers - [LeRobot Visualizer](https://huggingface.co/spaces/lerobot/visualize_dataset?path=%2FDaivdYuan%2Fhub-strap-wires-with-velcro-tape-lerobot) - [Neural Motion Visualizer](https://platform.neural-motion.org/open?repo=DaivdYuan%2Fhub-strap-wires-with-velcro-tape-lerobot) ## Overview This dataset is a conversion from an upstream robotics dataset into **LeRobot v3-compatible format**. It is intended to provide reproducible access in a unified schema. ## Source Dataset - Dataset ID: `hub-strap-wires-with-velcro-tape` - Project: `unknown` - Task: `Strap wires with velcro tape` - Upstream download/source URL: https://real.stanford.edu/maniwav/data/velcro_tape/replay_buffer.zarr.zip - Upstream project page: https://umi-data.github.io/ - Contact: liuzeyi@stanford.edu - Original paper: Data Scaling Laws in Imitation Learning for Robotic Manipulation (2024) - Notes: 193 demos; 1 envs; obs=Image,Proprio,Audio; actions=6 DoF End Effector,Parallel Gripper ## Dataset Format - Storage format: LeRobot v3 parquet-based dataset - Total episodes: 193 - Total frames: 151610 - FPS: 30 - Robot type: generic_robot - Observation feature: `observation.state` with shape `[7]` - Action feature: `action` with shape `[7]` - Video streams: present ## Video Modalities - `observation.images.camera0_rgb`: shape `[224, 224, 3]` - Frame rate: `30` FPS ## Conversion Notes - Target schema: LeRobot v3 - Conversion path: ingest -> preprocess -> zarr probe -> convert 1 episode -> upload - `action` fallback may be used when upstream `data/action` is missing. ## Dataset Structure `meta/info.json`: ```json { "codebase_version": "v3.0", "robot_type": "generic_robot", "total_episodes": 193, "total_frames": 151610, "fps": 30, "splits": {"train": "0:193"}, "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]}, "action": {"dtype": "float32", "shape": [7]} } } ``` ## Data Splits - `{'train': '0:193'}` ## Citation This dataset was created using [LeRobot](https://github.com/huggingface/lerobot). Please also cite the original upstream project and paper listed above. ## Attribution - Original dataset credit remains with upstream creators. - Conversion and curation contribution: - **Convertor: David Yuan, davidy02@stanford.edu, Stanford University** - **NMotion Initiative** (opensource community): *we make embodied data accessible*
license: 其他 tags: - 机器人学 - LeRobot - 机器人操作 - 未知 - NMotion倡议 --- # 威扣胶带捆扎导线任务 ## 可视化工具 - [LeRobot可视化工具](https://huggingface.co/spaces/lerobot/visualize_dataset?path=%2FDaivdYuan%2Fhub-strap-wires-with-velcro-tape-lerobot) - [神经运动可视化工具](https://platform.neural-motion.org/open?repo=DaivdYuan%2Fhub-strap-wires-with-velcro-tape-lerobot) ## 概述 本数据集是将上游机器人学数据集转换为**兼容LeRobot v3的格式**而来，旨在通过统一的架构提供可复现的数据访问途径。 ## 源数据集 - 数据集ID: `hub-strap-wires-with-velcro-tape` - 所属项目: `未知` - 任务: `威扣胶带（Velcro tape）捆扎导线` - 上游下载/源地址: https://real.stanford.edu/maniwav/data/velcro_tape/replay_buffer.zarr.zip - 上游项目页面: https://umi-data.github.io/ - 联系方式: liuzeyi@stanford.edu - 原始论文: 《机器人模仿学习中的数据缩放定律》（Data Scaling Laws in Imitation Learning for Robotic Manipulation，2024） - 说明: 共193段演示数据；1个实验环境；观测数据包含图像、本体感受、音频；动作数据为6自由度末端执行器、平行夹爪动作。 ## 数据集格式 - 存储格式: 基于Parquet（帕凯）的LeRobot v3数据集 - 总片段数: 193 - 总帧数: 151610 - 帧率: 30 - 机器人类型: 通用机器人 - 观测特征: `observation.state`，形状为`[7]` - 动作特征: `action`，形状为`[7]` - 包含视频流。 ## 视频模态 - `observation.images.camera0_rgb`: 形状为`[224, 224, 3]` - 帧率: `30` ## 转换说明 - 目标架构: LeRobot v3 - 转换流程: 数据摄入→预处理→Zarr（扎尔）探测→单片段转换→上传 - 当上游`data/action`字段缺失时，可使用`action`字段作为回退机制。 ## 数据集结构 `meta/info.json`: json { "codebase_version": "v3.0", "robot_type": "generic_robot", "total_episodes": 193, "total_frames": 151610, "fps": 30, "splits": {"train": "0:193"}, "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]}, "action": {"dtype": "float32", "shape": [7]} } } ## 数据划分 - `{"train": "0:193"}` ## 引用说明 本数据集基于[LeRobot](https://github.com/huggingface/lerobot)构建，请同时引用上述上游项目及原始论文。 ## 署名声明 - 原始数据集版权归上游创作者所有。 - 数据集转换与整理贡献: - **转换者: David Yuan，邮箱davidy02@stanford.edu，斯坦福大学** - **NMotion倡议（NMotion Initiative）**（开源社区）：*我们致力于让具身智能数据触手可及*
