gsethia08/part3-gather_roll_dice

该数据集由LeRobot项目创建，专注于机器人操作任务，特别是手部机器人的控制与交互。数据集包含1530个episodes，总计314316帧，覆盖39个不同任务，数据以parquet文件格式存储，视频文件为mp4格式，帧率为30fps。数据集特征丰富，包括观测数据（如第一人称视角的相机图像，分辨率为1080x1920，RGB三通道；手部状态信息，如左右手腕的位置、旋转和各个手指的坐标）、相机内外参数、动作向量、语言指令、时间戳、帧索引、episode索引等。这些数据可用于训练和评估机器人学习模型，支持强化学习、模仿学习等应用场景。数据集结构详细，通过meta/info.json文件提供了完整的元数据信息，便于研究人员使用和分析。This dataset is created by the LeRobot project, which focuses on robotic manipulation tasks, especially the control and interaction of dexterous robotic hands. It consists of 1530 episodes, totaling 314316 frames across 39 distinct tasks. The data is stored in Parquet file format, with video files in MP4 format at a frame rate of 30 FPS. The dataset includes rich modalities: observational data such as first-person RGB camera images with a resolution of 1080x1920, hand state information including the position, rotation of the left and right wrists, and coordinates of each finger, camera intrinsic and extrinsic parameters, action vectors, language instructions, timestamps, frame indices, and episode indices, among others. This data can be utilized to train and evaluate robotic learning models, supporting application scenarios such as reinforcement learning and imitation learning. The dataset has a detailed structure, with complete metadata provided via the meta/info.json file, facilitating researchers' usage and analysis.