taccap-g1-grab-usb-connector-insert-into-pc-port-0717
收藏资源简介：
该数据集是一个机器人学习数据集，使用LeRobot框架创建，专门针对双TacCap抓取器机器人（bi_taccap_gripper）设计。核心任务是抓取USB连接器并插入电脑端口（grab-usb-connector-insert-into-pc-port），包含60个完整情节（episodes），总计56742帧数据，帧率为30fps。数据集提供了多模态特征：动作数据包括左右机械臂的TCP（工具中心点）位置、旋转和夹爪位置，共20个浮点数值；状态观测数据与动作数据类似，也包含20个浮点数值，反映机器人的实时状态。此外，还包括触觉图像（左右触觉传感器的左右视图，分辨率为400x700像素，3通道）和视觉图像（左右腕部摄像头，分辨率为480x640像素，3通道），所有图像数据以视频格式存储，使用h264编码。数据集还包含时间戳、帧索引、情节索引等辅助信息，适用于机器人控制、强化学习和多模态感知研究。数据以Parquet文件格式存储，总数据文件大小为100MB，视频文件大小为200MB。
This dataset is a robot learning dataset created using the LeRobot framework, specifically designed for the bi_taccap_gripper robot. The core task is grab-usb-connector-insert-into-pc-port, containing 60 complete episodes with a total of 56,742 frames at a frame rate of 30fps. The dataset provides rich multimodal features: action data includes TCP (Tool Center Point) positions, rotations, and gripper positions for both left and right arms, totaling 20 floating-point values; state observation data is similar to action data, also containing 20 floating-point values reflecting the robots real-time status. Additionally, it includes tactile images (left and right views from left and right tactile sensors, with a resolution of 400x700 pixels, 3 channels) and visual images (left and right wrist cameras, with a resolution of 480x640 pixels, 3 channels), all stored in video format using h264 encoding. The dataset also contains auxiliary information such as timestamps, frame indices, and episode indices, making it suitable for robot control, reinforcement learning, and multimodal perception research. Data is stored in Parquet file format, with a total data file size of 100MB and video file size of 200MB.
数据集概览
- 数据集名称: taccap-g1-grab-usb-connector-insert-into-pc-port-0717
- 创建工具: LeRobot
- 许可证: Apache-2.0
- 机器人类型: bi_taccap_gripper
数据集规模
| 维度 | 数值 |
|---|---|
| 总片段数 (episodes) | 60 |
| 总帧数 (frames) | 56,742 |
| 任务数 (tasks) | 1 |
| 数据文件大小 | 100 MB |
| 视频文件大小 | 200 MB |
| 帧率 (FPS) | 30 |
数据集分割
- 训练集：片段 0 至 59（共 60 个片段）
特征说明
动作与状态特征 (action / observation.state)
- 数据类型: float32
- 维度: 20
- 包含: 左右机械臂 TCP 位置与姿态（x, y, z, r1~r6）、左右夹爪位置
图像观测 (observation.images)
| 相机视角 | 分辨率 | 格式 |
|---|---|---|
| left_tactile_left | 400×700×3 | h264 视频 |
| left_tactile_right | 400×700×3 | h264 视频 |
| left_wrist | 480×640×3 | h264 视频 |
| right_tactile_left | 400×700×3 | h264 视频 |
| right_tactile_right | 400×700×3 | h264 视频 |
| right_wrist | 480×640×3 | h264 视频 |
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]
引用信息
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} }
