leokswang/19122025-foldclo-12_lerobot_format
收藏资源简介:
该数据集由LeRobot创建,专注于机器人学领域,用于机器人控制和学习任务。数据集包含10个episodes,总计16699个帧,数据以parquet格式存储,每个episode被分成1个chunk,chunk大小为1000。机器人类型为yam,帧率为30fps,仅提供训练集(splits为train)。数据集特征包括:观测图像(顶部、左侧、右侧三个视角,每张图像为360x640x3的RGB图像)、观测状态(14维浮点数向量)、动作(14维浮点数向量),以及时间戳、帧索引、episode索引、索引和任务索引等元数据。这些数据可用于机器人行为模仿、强化学习等应用。
This dataset was created using LeRobot and focuses on the robotics domain, intended for robot control and learning tasks. It contains 10 episodes with a total of 16,699 frames, stored in parquet format. Each episode is divided into 1 chunk with a chunk size of 1000. The robot type is yam, the frame rate is 30 fps, and only a training set is provided (splits: train). The dataset features include: observation images (top, left, and right views, each image is an RGB image of size 360x640x3), observation state (a 14-dimensional float32 vector), action (a 14-dimensional float32 vector), and metadata such as timestamp, frame index, episode index, index, and task index. This data can be used for applications like robot behavior imitation and reinforcement learning.




