taccap-g1-Peel-arrange-close-earphone-Velcro-0729
收藏资源简介:
该数据集是一个机器人操作数据集,使用LeRobot框架创建,专门用于双TacCap夹爪机器人(bi_taccap_gripper)。数据集包含122个episodes,总计434278帧,涉及单个任务,帧率为30fps。数据以parquet格式存储,视频以mp4格式存储。数据集包括机器人动作数据(如左右夹爪的工具中心点位置、旋转参数和抓取器位置)和观察数据(如机器人状态和图像数据)。图像数据来自多个传感器:左右触觉传感器(左右触觉左和右触觉右,分辨率为400x700像素)和左右手腕摄像头(分辨率为480x640像素),所有图像均为彩色(3通道)。数据集适用于机器人学习、控制算法开发和触觉感知研究,支持训练和评估机器人操作任务。
This dataset is a robotic manipulation dataset developed with the LeRobot framework, tailored for the dual TacCap gripper robot (bi_taccap_gripper). It contains 122 episodes, totaling 434,278 frames, focusing on a single task with a frame rate of 30 fps. The dataset data is stored in Parquet format, while the related videos are stored in MP4 format. The dataset includes robotic action data (such as the tool center point positions, rotation parameters, and gripper positions of the left and right grippers) and observation data (such as robot states and image data). The image data is collected from multiple sensors: left and right tactile sensors (left tactile and right tactile, with a resolution of 400×700 pixels) and left and right wrist cameras (with a resolution of 480×640 pixels), all of which are 3-channel color images. This dataset is suitable for robotic learning, control algorithm development, and tactile perception research, and supports the training and evaluation of robotic manipulation tasks.
数据集总览
- 数据集名称: taccap-g1-Peel-arrange-close-earphone-Velcro-0729
- 创建工具: 使用 LeRobot 框架创建
- 许可证: Apache-2.0
- 任务类别: 机器人学 (robotics)
数据集规模
| 指标 | 数值 |
|---|---|
| 总片段数 (Episodes) | 122 |
| 总帧数 (Frames) | 434,278 |
| 总任务数 (Tasks) | 1 |
| 帧率 (FPS) | 30 |
| 数据文件大小 | 100 MB |
| 视频文件大小 | 200 MB |
数据拆分
- 训练集 (train): 全部 122 个片段(索引 0 至 122)
机器人类型
- 机器人型号:
bi_taccap_gripper(双臂触觉夹爪)
特征与观测空间
动作 (Action) 与观测状态 (Observation State)
- 数据类型: float32
- 维度: 20
- 各维度名称:
- 左臂 TCP 位置与姿态:
left_tcp.x,left_tcp.y,left_tcp.z,left_tcp.r1至left_tcp.r6 - 左夹爪位置:
left_gripper.pos - 右臂 TCP 位置与姿态:
right_tcp.x,right_tcp.y,right_tcp.z,right_tcp.r1至right_tcp.r6 - 右夹爪位置:
right_gripper.pos
- 左臂 TCP 位置与姿态:
视觉观测 (Camera Images)
| 摄像头名称 | 分辨率 (高×宽) | 通道数 | 编码格式 | 帧率 |
|---|---|---|---|---|
| left_tactile_left (左触觉左) | 400×700 | 3 | H.264 | 30 |
| left_tactile_right (左触觉右) | 400×700 | 3 | H.264 | 30 |
| left_wrist (左手腕) | 480×640 | 3 | H.264 | 30 |
| right_tactile_left (右触觉左) | 400×700 | 3 | H.264 | 30 |
| right_tactile_right (右触觉右) | 400×700 | 3 | H.264 | 30 |
| right_wrist (右手腕) | 480×640 | 3 | H.264 | 30 |
其他特征
timestamp: float32, 形状 [1]frame_index: int64, 形状 [1]episode_index: int64, 形状 [1]index: int64, 形状 [1]task_index: int64, 形状 [1]
数据文件结构
- 数据路径:
data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet - 视频路径:
videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4 - 分块大小 (chunks_size): 1000
- 元数据文件:
meta/info.json
引用格式 (BibTeX)
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} }



