aaawangge/rollout_act_so101_precision-pen-placement

Name: aaawangge/rollout_act_so101_precision-pen-placement
Creator: aaawangge
Published: 2026-05-29 06:51:14
License: 暂无描述

Hugging Face2026-05-29 更新2026-05-31 收录

下载链接：

https://hf-mirror.com/datasets/aaawangge/rollout_act_so101_precision-pen-placement

下载链接

链接失效反馈

官方服务：

资源简介：

--- license: apache-2.0 task_categories: - robotics tags: - LeRobot configs: - config_name: default data_files: data/*/*.parquet --- This dataset was created using [LeRobot](https://github.com/huggingface/lerobot). <a class="flex" href="https://huggingface.co/spaces/lerobot/visualize_dataset?path=aaawangge/rollout_act_so101_precision-pen-placement_20260529_143834"> <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", "fps": 30, "features": { "action": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6 ] }, "observation.state": { "dtype": "float32", "names": [ "shoulder_pan.pos", "shoulder_lift.pos", "elbow_flex.pos", "wrist_flex.pos", "wrist_roll.pos", "gripper.pos" ], "shape": [ 6 ] }, "observation.images.top": { "dtype": "video", "shape": [ 480, 640, 3 ], "names": [ "height", "width", "channels" ], "info": { "video.height": 480, "video.width": 640, "video.codec": "av1", "video.pix_fmt": "yuv420p", "video.is_depth_map": false, "video.fps": 30, "video.channels": 3, "has_audio": false, "video.g": 2, "video.crf": 30, "video.preset": 12, "video.fast_decode": 0, "video.video_backend": "pyav", "video.extra_options": {} } }, "observation.images.front": { "dtype": "video", "shape": [ 480, 640, 3 ], "names": [ "height", "width", "channels" ], "info": { "video.height": 480, "video.width": 640, "video.codec": "av1", "video.pix_fmt": "yuv420p", "video.is_depth_map": false, "video.fps": 30, "video.channels": 3, "has_audio": false, "video.g": 2, "video.crf": 30, "video.preset": 12, "video.fast_decode": 0, "video.video_backend": "pyav", "video.extra_options": {} } }, "observation.images.right": { "dtype": "video", "shape": [ 480, 640, 3 ], "names": [ "height", "width", "channels" ], "info": { "video.height": 480, "video.width": 640, "video.codec": "av1", "video.pix_fmt": "yuv420p", "video.is_depth_map": false, "video.fps": 30, "video.channels": 3, "has_audio": false, "video.g": 2, "video.crf": 30, "video.preset": 12, "video.fast_decode": 0, "video.video_backend": "pyav", "video.extra_options": {} } }, "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 } }, "total_episodes": 20, "total_frames": 12000, "total_tasks": 1, "chunks_size": 1000, "data_files_size_in_mb": 100, "video_files_size_in_mb": 200, "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", "robot_type": "so_follower", "splits": { "train": "0:20" } } ``` ## Citation **BibTeX:** ```bibtex [More Information Needed] ```

This dataset is designed for robotics tasks and was created using the LeRobot tool. It consists of 20 episodes and 12,000 frames, stored in parquet format with accompanying video files. The features include robot actions (e.g., shoulder pan position, shoulder lift position, elbow flexion position, wrist flexion position, wrist roll position, gripper position), observation states (including joint positions and images from top, front, and right cameras, with image resolution of 480x640, 3 channels, video codec av1, frame rate 30fps), as well as timestamp, frame index, episode index, index, and task index. The robot type is so_follower, with a sampling frequency of 30fps. The total dataset size is 300MB (100MB for data files and 200MB for video files), and it is split into a training set (episodes 0 to 20).

提供机构：

aaawangge

搜集汇总

数据集介绍

构建方式

该数据集基于LeRobot框架构建，专注于精密笔放置任务，由so_follower机器人执行20个操作回合（episodes）采集而成，共计12,000帧数据。数据以parquet格式存储，并伴有三种视角（顶部、前部、右侧）的高清视频记录，视频编码采用AV1，分辨率为480×640，帧率30 FPS。数据集结构遵循v3.0规范，将数据与视频文件分块存储，便于高效加载与处理。

特点

数据集的核心特点在于其多模态与高精度特性：同时包含6维关节位置指令（action）与观测状态（observation.state），覆盖机器人肩部、肘部、腕部及夹爪的完整运动空间。三类视觉观测（top、front、right）提供了多角度环境感知能力，适合训练模仿学习与强化学习模型。所有数据来源于真实机器人操作，保证了行为的物理可行性与任务相关性。

使用方法

用户可通过LeRobot库的DataPipeline轻松加载该数据集，利用提供的train拆分（0-20个episodes）进行模型训练。数据包含连续的帧索引与时间戳，便于时序建模。建议将6维动作向量与视觉输入结合，用于训练策略网络。数据集支持直接可视化浏览（HuggingFace Space），方便快速检查样本质量与任务特性。

背景与挑战

背景概述

该数据集由aaawangge团队于2025年创建，依托LeRobot框架构建，核心聚焦于精密笔放置任务。在机器人操作领域，高精度物体放置是衡量机械臂灵巧性与控制算法性能的关键基准，涉及视觉感知、运动规划与力反馈的协同优化。该数据集通过采集so_follower型机械臂在六自由度空间中的动作序列与多视角视觉信息，为模仿学习与强化学习算法提供了标准化的训练与评估平台。其发布推动了机器人精细操作技能的自动化学习研究，尤其对工业装配、医疗手术等需要亚毫米级精度场景具有显著参考价值。

当前挑战

数据集面临的挑战首先体现在领域问题层面：精密笔放置任务要求机械臂在厘米级误差范围内完成姿态调整，现有算法常因视觉模糊、关节摩擦或运动学模型偏差导致失败。构建过程中，多视角视频采集面临光照变化与遮挡问题，30FPS的采样率需平衡运动流畅性与数据存储成本；20个episode的规模可能无法覆盖复杂场景的多样性，而动作与状态空间均依赖单一机械臂型号，限制了跨平台泛化能力。此外，数据标注依赖遥操作演示，人工成本高昂且一致性难以保障，成为规模化扩展的主要瓶颈。

常用场景

经典使用场景

在机器人学习与精细操作领域，rollout_act_so101_precision-pen-placement数据集专为研究高精度物体放置任务而生。其核心价值在于提供了20个完整轨迹片段，包含多视角视觉观测（顶部、前方、右侧相机，640×480分辨率，30fps）与6自由度关节状态数据（肩部、肘部、腕部及夹爪），可支持模仿学习与行为克隆模型的训练与评估。经典使用场景涵盖基于视觉的机器人精确抓取-放置管线构建，研究人员常利用该数据集训练策略网络，使机械臂从示教轨迹中学习从初始姿态到精准放置笔类物体的复杂映射关系。

衍生相关工作

围绕该数据集的仿射属性，衍生出多项开创性工作。首先，基于其多视角图像序列与动作标签的共轭结构，研究者开发了融合时空注意力机制的视觉预训练模型（如基于VideoMAE的机器人操作变体）。其次，以该数据集为评估基准，催生了针对“稀疏奖励下长程任务”的隐式规划算法，通过学习潜在动作残差显著提升放置成功率。此外，部分工作基于其6D状态空间的拓扑特性，引入能量函数约束实现轨迹平滑与奇异点规避，推动了任务与动作解耦的模块化策略发展。

数据集最近研究