eval_depthDT200_unseen_draw
收藏资源简介:
该数据集是使用LeRobot创建的机器人数据集,专门用于机器人学习和控制任务。数据集包含10个episodes,总计2371帧,帧率为30fps。数据以parquet格式存储,视频以mp4格式存储。机器人类型为so_follower,特征包括动作(6个关节位置:肩部平移、肩部升降、肘部弯曲、腕部弯曲、腕部旋转、夹爪位置)、观测状态(6个关节位置)、两个摄像头图像(侧视和腕部摄像头,分辨率480x640,RGB通道,无音频),以及时间戳、帧索引、episode索引、索引和任务索引等元数据。数据集适用于训练和评估机器人控制模型,支持从多模态观测中学习动作策略。
This dataset is a robot dataset created using LeRobot, specifically designed for robot learning and control tasks. It contains 10 episodes, totaling 2371 frames, with a frame rate of 30fps. The data is stored in parquet format, and videos are stored in mp4 format. The robot type is so_follower, and features include actions (6 joint positions: shoulder translation, shoulder lift, elbow bend, wrist bend, wrist rotation, gripper position), observation states (6 joint positions), two camera images (side view and wrist camera, resolution 480x640, RGB channels, no audio), as well as metadata such as timestamps, frame index, episode index, index, and task index. The dataset is suitable for training and evaluating robot control models, supporting the learning of action policies from multimodal observations.
数据集概述
- 数据集名称: eval_depthDT200_unseen_draw
- 许可证: Apache-2.0
- 任务类别: 机器人学 (Robotics)
- 创建工具: LeRobot
数据集规模
- 总片段数 (Episodes): 10
- 总帧数 (Frames): 2371
- 总任务数 (Tasks): 1
- 帧率 (FPS): 30
- 数据文件大小: 约 100 MB (Parquet 格式)
- 视频文件大小: 约 200 MB (MP4 格式)
- 分割: 训练集包含全部 10 个片段 (train: 0:10)
机器人类型
- 机器人: so_follower
数据特征
动作 (action) 与 观测状态 (observation.state):
- 数据类型: float32
- 维度: 6 (包含肩关节、肘关节、腕关节及夹爪的位置)
shoulder_pan.posshoulder_lift.poselbow_flex.poswrist_flex.poswrist_roll.posgripper.pos
观测图像 (observation.images):
- 侧方摄像头 (side): 视频格式 (AV1 编码), 分辨率 480x640, 3 通道, 30 FPS, 非深度图
- 腕部摄像头 (wrist): 视频格式 (AV1 编码), 分辨率 480x640, 3 通道, 30 FPS, 非深度图
时间戳与索引:
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]
数据格式
- 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
可视化
- 可通过 LeRobot 可视化空间 交互查看数据集。



