taccap-g1-scoop-coffee-granules-0707
收藏Hugging Face2026-07-07 更新2026-07-07 收录
下载链接:
https://huggingface.co/datasets/TacVerse/taccap-g1-scoop-coffee-granules-0707
下载链接
链接失效反馈官方服务:
资源简介:
该数据集由LeRobot创建,专注于机器人学任务,具体涉及使用双机械手夹爪系统(bi_taccap_gripper)执行舀取咖啡颗粒的操作。数据集包含30个完整的情节(episodes),总计23747帧数据,覆盖单一任务。数据内容包括动作信息(如左右机械手的TCP位置、旋转和夹爪位置)、状态观察(与动作类似的结构)、多视角图像观察(包括左腕和右腕摄像头图像,以及左右触觉传感器的视频数据),以及时间戳、帧索引、情节索引等元数据。图像和视频数据以高分辨率提供(例如腕部摄像头为480x640,触觉传感器为400x700),帧率为30fps。数据集以Apache 2.0许可证发布,适用于机器人控制、模仿学习或强化学习研究。
This dataset is created by LeRobot and focuses on robotics tasks, specifically involving the use of a dual-arm gripper system (bi_taccap_gripper) to perform the operation of scooping coffee granules. It contains 30 complete episodes, totaling 23,747 frames of data, covering a single task. The data includes action information (such as TCP positions, rotations, and gripper positions for both left and right arms), state observations (structured similarly to actions), multi-view image observations (including left and right wrist camera images, as well as video data from left and right tactile sensors), and metadata such as timestamps, frame indices, and episode indices. Image and video data are provided in high resolution (e.g., 480x640 for wrist cameras and 400x700 for tactile sensors) at a frame rate of 30fps. The dataset is released under the Apache 2.0 license and is suitable for research in robot control, imitation learning, or reinforcement learning.
提供机构:
TacVerse创建时间:
2026-07-07
原始信息汇总
数据集总览
- 数据集名称: TacCap G1 Scoop Coffee Granules 0707
- 许可证: Apache-2.0
- 任务类别: 机器人学 (robotics)
- 数据集标签: LeRobot
- 创建工具: 该数据集通过 LeRobot 创建
数据集规模
- 总片段数: 30
- 总帧数: 23747
- 任务数量: 1
- 帧率: 30 FPS
- 数据文件大小: 100 MB
- 视频文件大小: 200 MB
- 数据分割: 全部30个片段用于训练
机器人类型
- 机器人型号: bi_taccap_gripper
数据集结构
- 数据文件路径:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频文件路径:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4 - 分块大小: 1000
特征说明
动作 (action) 与观测状态 (observation.state)
两者均为20维向量,包含以下参数:
- 左/右TCP位置与姿态:包含x, y, z坐标以及r1至r6的方向参数
- 左/右夹爪位置:left_gripper.pos, right_gripper.pos
观测图像 (observation.images)
数据集包含6个摄像头视角,均为视频类型,30 FPS,H.264编码,无音频:
| 图像键名 | 分辨率 (高×宽) | 通道数 |
|---|---|---|
| left_tactile_left | 400 × 700 | 3 (RGB) |
| left_tactile_right | 400 × 700 | 3 (RGB) |
| left_wrist | 480 × 640 | 3 (RGB) |
| right_tactile_left | 400 × 700 | 3 (RGB) |
| right_tactile_right | 400 × 700 | 3 (RGB) |
| right_wrist | 480 × 640 | 3 (RGB) |
其他特征
- timestamp: 时间戳,float32,形状为[1]
- frame_index: 帧索引,int64,形状为[1]
- episode_index: 片段索引,int64,形状为[1]
- index: 索引,int64,形状为[1]
- task_index: 任务索引,int64,形状为[1]
引用
bibtex @misc{xense-taccap-lerobot, author = {XenseRobotics Team}, title = {LeRobot-Xense: LeRobot with Xense Tactile Robotics Support}, howpublished = {url{https://github.com/Vertax42/xense-taccap-lerobot}}, year = {2026} }



