five

dillonlyr04/eval_final_k20_vf

收藏
Hugging Face2026-05-27 更新2026-05-31 收录
下载链接:
https://hf-mirror.com/datasets/dillonlyr04/eval_final_k20_vf
下载链接
链接失效反馈
官方服务:
资源简介:
这是一个使用LeRobot创建的机器人学数据集,专门针对双臂机器人(bi_so100_follower)。数据集包含1个总情节、841个总帧数和1个总任务,数据以Parquet文件格式存储,并附带视频文件。数据特征包括动作(action)和观测状态(observation.state),其中动作和状态数据都包含左右肩、肘、腕和夹爪的12个位置参数(浮点型)。此外,观测部分还包括来自顶部全局视角、左臂和右臂的三个图像视频流,每个视频的分辨率为720x1280,3个通道,帧率为30fps,使用h264编码。其他元数据包括时间戳、帧索引、情节索引、索引和任务索引。数据集适用于机器人控制、模仿学习或强化学习任务,并遵循Apache 2.0许可证。

This dataset was created using LeRobot and focuses on robotics, specifically for a bi_so100_follower dual-arm robot. It contains 1 total episode, 841 total frames, and 1 total task, with data stored in Parquet files and accompanied by video files. The features include action and observation.state, both comprising 12 position parameters (float32) for left and right shoulder, elbow, wrist, and gripper. Additionally, observations include three image video streams from top_global, left_arm, and right_arm perspectives, each with a resolution of 720x1280, 3 channels, 30fps, and h264 codec. Other metadata includes timestamp, frame_index, episode_index, index, and task_index. The dataset is suitable for robotics control, imitation learning, or reinforcement learning tasks and is licensed under Apache 2.0.
提供机构:
dillonlyr04
二维码
社区交流群
二维码
科研交流群
商业服务