资源简介：
该数据集使用LeRobot创建，专注于机器人黄瓜抓取任务（基于autoresearch-000109-cucumber-grab上下文）。数据集包含机器人状态观测、多视角视频观测和动作数据：状态观测为7维浮点型向量，表示关节位置（包括肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转、夹爪位置和倾斜位置）；视频观测来自三个视角（顶部、腕部和基座），每个视频分辨率为480x640x3，帧率为20fps，使用AV1编码；动作为7维浮点型向量，对应关节位置控制。此外，数据集还包含时间戳、帧索引、回合索引、索引和任务索引等元数据。数据集总共有1个回合、147帧和1个任务，数据以parquet文件存储，视频以mp4文件存储，总数据大小约100MB，视频大小约200MB，机器人类型为vibeboard_follower_tilt。该数据集适用于机器人学习、视觉伺服和动作预测等研究领域。
This dataset is created using LeRobot and focuses on a robotic cucumber grasping task (based on the autoresearch-000109-cucumber-grab context). It includes robot state observations, multi-view video observations, and action data: state observations are 7-dimensional floating-point vectors representing joint positions (including shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist rotation, gripper position, and tilt position); video observations come from three perspectives (top, wrist, and base), each with a resolution of 480x640x3, a frame rate of 20fps, and encoded using AV1; actions are 7-dimensional floating-point vectors corresponding to joint position control. Additionally, the dataset contains metadata such as timestamps, frame indices, episode indices, indices, and task indices. The dataset consists of 1 episode, 147 frames, and 1 task, with data stored in parquet files and videos in mp4 files. The total data size is approximately 100MB, and the video size is about 200MB. The robot type is vibeboard_follower_tilt. This dataset is suitable for research in robotics learning, visual servoing, and action prediction.
