eval2_bgr_g_eval1prompt
收藏资源简介:
该数据集是一个机器人学习数据集,使用LeRobot工具创建。数据集包含4个完整的情节(episodes),总计2400帧数据,对应1个任务。数据以30fps的帧率采集,总数据文件大小约为100MB,视频文件大小约为200MB。数据集结构包含多个特征字段:动作(action)和状态观测(observation.state)均为6维浮点向量,分别对应机械臂的肩部平移、肩部抬升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置;观测图像(observation.images.front)为来自前置摄像头的视频数据,分辨率480x640,3通道彩色,采用AV1编解码器;此外还包括时间戳、帧索引、情节索引、全局索引和任务索引等元数据字段。所有数据均存储在Parquet文件格式中,视频数据存储在MP4文件中。数据集适用于机器人模仿学习、强化学习或行为克隆等任务。
This dataset is a robot learning dataset created using the LeRobot tool. It contains 4 complete episodes, totaling 2400 frames of data, corresponding to 1 task. The data is collected at a frame rate of 30fps, with a total data file size of approximately 100MB and a video file size of about 200MB. The dataset structure includes multiple feature fields: action and observation.state are both 6-dimensional floating-point vectors, corresponding to the robot arms shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist rotation, and gripper position, respectively; observation.images.front is video data from a front-facing camera, with a resolution of 480x640, 3-channel color, using the AV1 codec; additionally, it includes metadata fields such as timestamp, frame index, episode index, global index, and task index. All data is stored in Parquet file format, with video data stored in MP4 files. The dataset is suitable for tasks such as robot imitation learning, reinforcement learning, or behavior cloning.




