Trainings/20260528record-test_20260528_163329
收藏资源简介:
该数据集是一个用于机器人学习的数据集,专门针对机器人控制任务。它包含5个完整的情节(episodes),总计316帧数据,数据以parquet文件格式存储,同时提供mp4格式的视频文件。数据集的特征包括动作(action)和观察(observation),其中动作表示机器人的关节位置控制,涵盖肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置等6个维度;观察包括机器人的状态(同样为6个关节位置)以及来自侧视和俯视摄像头的图像数据,图像分辨率为480x640,帧率为30fps。数据集适用于训练和评估机器人强化学习或模仿学习模型,支持多模态输入(如状态和视觉信息)。数据集的元信息指示其使用LeRobot工具包创建,机器人类型为so_follower,并采用Apache 2.0许可证。
This dataset is a robotics learning dataset designed for robot control tasks. It contains 5 complete episodes with a total of 316 frames, stored in parquet format for data and mp4 format for videos. The dataset features include action and observation, where action represents robot joint position control across 6 dimensions: shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper position. Observation comprises the robot state (also 6 joint positions) and image data from side and up cameras, with a resolution of 480x640 and a frame rate of 30fps. The dataset is suitable for training and evaluating robot reinforcement learning or imitation learning models, supporting multimodal inputs such as state and visual information. Metadata indicates it was created using the LeRobot toolkit, with robot type so_follower, and is licensed under Apache 2.0.



