five

brandonyang/droid_overlay_v2

收藏
Hugging Face2026-05-25 更新2026-05-31 收录
下载链接:
https://hf-mirror.com/datasets/brandonyang/droid_overlay_v2
下载链接
链接失效反馈
官方服务:
资源简介:
该数据集由LeRobot创建,专注于机器人技术领域,主要用于轨迹覆盖和多剧集任务。数据集包含5289个总剧集、862822个总帧和3076个总任务,数据以parquet文件格式存储,分为6个块,每个块大小为1000。帧率为15fps,机器人类型为panda。数据集特征包括:三个图像特征(exterior_image_1_left、exterior_image_2_left、wrist_image_left),每个图像的分辨率为180x320x3;关节位置(7维浮点数组);夹持器位置(1维浮点数组);动作(8维浮点数组);以及时间戳、帧索引、剧集索引、索引和任务索引等元数据。所有数据仅用于训练集,视频路径存在但总视频数为0。

许可证:Apache-2.0 任务类别:机器人学 标签:LeRobot、droid、轨迹叠加(trajectory-overlay)、多回合(multi-episode)、完整(full)、笔记本风格(notebook-style) 配置项: - 配置名称:default 数据文件:data/*/*.parquet 本数据集基于[LeRobot](https://github.com/huggingface/lerobot)构建。 ## 数据集说明 - **主页**:[更多信息待补充] - **论文**:[更多信息待补充] - **许可证**:Apache-2.0 ## 数据集结构 [meta/info.json](meta/info.json): json { "codebase_version": "v2.1", "robot_type": "panda", "total_episodes": 5289, "total_frames": 862822, "total_tasks": 3076, "total_videos": 0, "total_chunks": 6, "chunks_size": 1000, "fps": 15, "splits": { "train": "0:5289" }, "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "video_path": "videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4", "features": { "exterior_image_1_left": { "dtype": "image", "shape": [ 180, 320, 3 ], "names": [ "height", "width", "channel" ] }, "exterior_image_2_left": { "dtype": "image", "shape": [ 180, 320, 3 ], "names": [ "height", "width", "channel" ] }, "wrist_image_left": { "dtype": "image", "shape": [ 180, 320, 3 ], "names": [ "height", "width", "channel" ] }, "joint_position": { "dtype": "float32", "shape": [ 7 ], "names": [ "joint_position" ] }, "gripper_position": { "dtype": "float32", "shape": [ 1 ], "names": [ "gripper_position" ] }, "actions": { "dtype": "float32", "shape": [ 8 ], "names": [ "actions" ] }, "timestamp": { "dtype": "float32", "shape": [ 1 ], "names": null }, "frame_index": { "dtype": "int64", "shape": [ 1 ], "names": null }, "episode_index": { "dtype": "int64", "shape": [ 1 ], "names": null }, "index": { "dtype": "int64", "shape": [ 1 ], "names": null }, "task_index": { "dtype": "int64", "shape": [ 1 ], "names": null } } } ## 引用 **BibTeX格式**: bibtex [更多信息待补充]
提供机构:
brandonyang
二维码
社区交流群
二维码
科研交流群
商业服务