KoukiHagiwara/slide_push_pull_hybrid_250ep_augmented
收藏资源简介:
该数据集是通过LeRobot创建的机器人技术数据集,用于训练和评估机器人动作。数据集包含250个训练集,总帧数为262,000帧,涉及2个任务。数据采集自so_follower机器人类型,包括动作数据(如肩部、肘部、手腕和夹爪的位置)、观察状态(与动作数据类似)以及来自顶部摄像头和手腕摄像头的图像观察(视频格式,分辨率480x640,30fps)。数据集还包含时间戳、帧索引、剧集索引等元数据。在创建过程中,进行了50次横向动作(暗环境)、50次横向动作(明亮环境)、100次斜向动作,每个动作都进行了两次往返,教示动作范围在x方向-60mm到0mm,y方向50mm内。此外,使用OpenCV对视频亮度进行了调整。
This dataset was created using LeRobot for robotics applications. It consists of 250 episodes with a total of 262,000 frames and covers 2 tasks. The data is based on the so_follower robot type and includes action data (e.g., positions of shoulder, elbow, wrist, and gripper), observation states (similar to action data), and image observations from a top camera and a wrist camera (in video format with 480x640 resolution, 30fps). The dataset also contains metadata such as timestamps, frame indices, episode indices, etc. During creation, 50 lateral movements (in dark conditions), 50 lateral movements (in bright conditions), and 100 diagonal movements were performed, each with two round-trip actions, within a teaching range of -60mm < x < 0 and 50mm in the y-direction. Additionally, video brightness was adjusted using OpenCV.




