ecappiell/weighted_boxes_experiment_20260529_093115
收藏资源简介:
该数据集由LeRobot创建,是一个机器人学数据集,专门用于so_follower类型机器人的研究。数据集包含5个总片段和3492个总帧数,数据以parquet格式存储,视频编码为av1。数据集的特征包括:动作数据(6维浮点数组,表示6个关节的位置)、观察状态数据(111维浮点数组,涵盖关节位置、原始电流、负载、硬件速度、速度、原始扭矩、签名电流和扭矩、滤波速度、估计外部扭矩、模型扭矩(如电机、重力、摩擦、科里奥利、惯性扭矩)以及工具中心点(TCP)的力和力矩、HPI夹爪的外部扭矩和位置信息)、两个图像观测(顶部和角落视角的视频,分辨率480x640,3通道,30fps),以及其他元数据如时间戳、帧索引、片段索引、索引和任务索引。数据集结构支持训练分割,总数据文件大小为100MB,视频文件大小为200MB。
This dataset was created using LeRobot and is a robotics dataset specifically for the so_follower robot type. It contains a total of 5 episodes and 3492 frames, with data stored in parquet format and videos encoded in av1. The dataset features include: action data (a 6-dimensional float32 array representing positions of 6 joints), observation state data (a 111-dimensional float32 array covering joint positions, raw current, load, hardware velocity, velocity, raw torque, signed current and torque, filtered velocity, estimated external torque, model torques (e.g., motor, gravity, friction, coriolis, inertial torque), as well as tool center point (TCP) forces and torques, HPI gripper external torque and position information), two image observations (videos from top and corner perspectives, resolution 480x640, 3 channels, 30fps), and other metadata such as timestamp, frame index, episode index, index, and task index. The dataset structure supports a train split, with total data file size of 100MB and video file size of 200MB.




