ethanCSL/eval_steering_ours_high_6_fixed

该数据集是使用LeRobot创建的机器人数据集，专为机器人控制任务设计。数据集包含一个名为“koch_follower”的机器人类型，数据以Parquet文件格式存储，总大小为100MB（数据文件）和200MB（视频文件），帧率为30fps。数据特征包括动作（action）和观察（observation）两部分：动作由6个关节位置（如肩部平移、肩部提升、肘部弯曲、手腕弯曲、手腕滚动和夹爪位置）组成，数据类型为float32；观察包括状态（与动作相同的6个关节位置）和来自三个相机的图像（camera1、camera2、camera3），每个相机图像分辨率为480x640像素、3个通道（RGB），以视频格式存储。此外，数据集还包含时间戳、帧索引、剧集索引、索引和任务索引等元数据，用于跟踪数据序列。数据集结构支持分块存储，每个块大小为1000条记录，适用于机器人学习和模仿学习任务。

This dataset is a robotics dataset created using LeRobot, designed for robot control tasks. It features a robot type named koch_follower, with data stored in Parquet format, total sizes of 100MB for data files and 200MB for video files, and a frame rate of 30fps. The dataset includes features for action and observation: actions consist of 6 joint positions (e.g., shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper positions) with a float32 data type; observations include state (same 6 joint positions as actions) and images from three cameras (camera1, camera2, camera3), each with a resolution of 480x640 pixels and 3 channels (RGB), stored in video format. Additionally, metadata such as timestamp, frame index, episode index, index, and task index are provided to track data sequences. The dataset structure supports chunked storage with a chunk size of 1000 records, making it suitable for robot learning and imitation learning applications.