berkeley_gnm_recon

This dataset was created using [LeRobot](https://github.com/huggingface/lerobot). <a class="flex" href="https://huggingface.co/spaces/lerobot/visualize_dataset?path=lerobot/berkeley_gnm_recon"> <img class="block dark:hidden" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/visualize-this-dataset-xl.svg"/> <img class="hidden dark:block" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/visualize-this-dataset-xl-dark.svg"/> </a> ## Dataset Description - **Homepage:** [More Information Needed] - **Paper:** [More Information Needed] - **License:** apache-2.0 ## Dataset Structure [meta/info.json](meta/info.json): ```json { "codebase_version": "v3.0", "robot_type": "unknown", "total_episodes": 11834, "total_frames": 610907, "total_tasks": 1, "chunks_size": 1000, "fps": 3, "splits": { "train": "0:11834" }, "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet", "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4", "features": { "observation.images.image": { "dtype": "video", "shape": [ 120, 160, 3 ], "names": [ "height", "width", "channel" ], "video_info": { "video.fps": 3.0, "video.codec": "av1", "video.pix_fmt": "yuv420p", "video.is_depth_map": false, "has_audio": false } }, "observation.state": { "dtype": "float32", "shape": [ 9 ], "names": { "motors": [ "motor_0", "motor_1", "motor_2", "motor_3", "motor_4", "motor_5", "motor_6", "motor_7", "motor_8" ] }, "fps": 3 }, "action": { "dtype": "float32", "shape": [ 7 ], "names": { "motors": [ "motor_0", "motor_1", "motor_2", "motor_3", "motor_4", "motor_5", "motor_6" ] }, "fps": 3 }, "timestamp": { "dtype": "float32", "shape": [ 1 ], "names": null, "fps": 3 }, "episode_index": { "dtype": "int64", "shape": [ 1 ], "names": null, "fps": 3 }, "frame_index": { "dtype": "int64", "shape": [ 1 ], "names": null, "fps": 3 }, "next.reward": { "dtype": "float32", "shape": [ 1 ], "names": null, "fps": 3 }, "next.done": { "dtype": "bool", "shape": [ 1 ], "names": null, "fps": 3 }, "index": { "dtype": "int64", "shape": [ 1 ], "names": null, "fps": 3 }, "task_index": { "dtype": "int64", "shape": [ 1 ], "names": null, "fps": 3 } }, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200 } ``` ## Citation **BibTeX:** ```bibtex [More Information Needed] ```本数据集基于[LeRobot](https://github.com/huggingface/lerobot)开发构建。 ## 数据集说明 - **项目主页**：https://sites.google.com/view/recon-robot - **相关论文**：https://arxiv.org/abs/2104.05859 - **授权协议**：MIT许可证 ## 数据集结构 文件`meta/info.json`内容如下： json { "代码库版本": "v2.0", "机器人类型": "未知", "总回合数": 11834, "总帧数": 610907, "总任务数": 1, "总视频数": 11834, "总分块数": 12, "单分块标准大小": 1000, "帧率": 3, "数据集划分": { "训练集": "0:11834" }, "数据文件路径格式": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", "视频文件路径格式": "videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4", "数据特征": { "观测.图像": { "数据类型": "视频", "形状": [120, 160, 3], "维度名称": ["高度", "宽度", "通道数"], "视频信息": { "视频帧率": 3.0, "视频编码格式": "av1", "视频像素格式": "yuv420p", "是否为深度图": false, "是否包含音频": false } }, "语言指令": { "数据类型": "字符串", "形状": [1], "维度名称": null }, "观测.状态": { "数据类型": "float32", "形状": [9], "维度名称": { "电机": ["motor_0", "motor_1", "motor_2", "motor_3", "motor_4", "motor_5", "motor_6", "motor_7", "motor_8"] } }, "动作": { "数据类型": "float32", "形状": [7], "维度名称": { "电机": ["motor_0", "motor_1", "motor_2", "motor_3", "motor_4", "motor_5", "motor_6"] } }, "时间戳": { "数据类型": "float32", "形状": [1], "维度名称": null }, "回合索引": { "数据类型": "int64", "形状": [1], "维度名称": null }, "帧索引": { "数据类型": "int64", "形状": [1], "维度名称": null }, "下一时刻奖励": { "数据类型": "float32", "形状": [1], "维度名称": null }, "回合结束标记": { "数据类型": "bool", "形状": [1], "维度名称": null }, "全局索引": { "数据类型": "int64", "形状": [1], "维度名称": null }, "任务索引": { "数据类型": "int64", "形状": [1], "维度名称": null } } } ## 引用 **BibTeX 格式引用**： bibtex @inproceedings{shah2021rapid, title={Rapid Exploration for Open-World Navigation with Latent Goal Models}, author={Dhruv Shah and Benjamin Eysenbach and Nicholas Rhinehart and Sergey Levine}, booktitle={第五届机器人学习年度会议}, year={2021}, url={https://openreview.net/forum?id=d_SWJhyKfVw} }

搜集汇总

背景与挑战

背景概述

该数据集由LeRobot创建，采用Apache 2.0许可证，包含11834个episodes和610907帧数据，涵盖视频观察、机器人状态和动作等结构化特征。数据以parquet和mp4格式存储，用于机器人任务相关的训练与分析。