产业观察 [table_Header]2025.12.02 【具身智能产业动态】松延动力完成近 2 亿元 Pre- 产业研究中心 B 轮融资，灵心巧手完成数亿元 A 加轮融资 摘要：产业最新趋势跟踪，点评产业最新风向 [Table_Summary] 具身机器人产业动态 国内规模最大的全尺寸人形机器人数据集 LET 数据集开源发布 工信部人形机器人标准化技术委员会名单公布，王兴兴、稚晖君入选 极佳视界发布轮式人形机器人 Maker H01 智元机器人上线灵心平台，用户可通过简易操作实现个性化机器人配置 星动纪元发布物流仓储解决方案，全球首次端到端 VLA 在物流仓储真实应用 智能汽车产业动态 汽车销量榜：10月新势力交付排行榜前三名为零跑（70289辆）、鸿蒙智行（68216 辆）、小鹏（42013 辆）。 小马智行发布 Q3 财报，城市级 UE 转正，Robotaxi 千台目标将提前完成，2026 年规模扩至 3000 辆 小马智行与阳光出行建立战略合作，全面推进自动驾驶出行业务发展 九识智能与广州高新兴机器人战略合作，共推无人巡检创新应用 西井科技与和记阿曼苏哈尔港口签约，15 台 E-Truck 即将进驻 ...

Core Insights - Leju Intelligent has donated its LET dataset to the OpenLoong open-source community, marking a significant step in the development of humanoid robots in China [1][4] - The LET dataset is a comprehensive collection of real-world data, exceeding 60,000 minutes, covering various operational scenarios across multiple industries [2][3] Group 1: Dataset Characteristics - The LET dataset is constructed to represent real operational scenarios for full-sized humanoid robots, encompassing industrial, commercial retail, and daily life environments [2] - It includes 31 tasks and 117 atomic skills, forming a clear task system that supports multi-scenario, multi-step, and multi-objective learning and reasoning for robots [2] Group 2: Industry Challenges and Solutions - The humanoid robotics industry faces challenges such as fragmented data sources and inconsistent formats, which hinder data quality and collaborative efficiency [3] - The donation of the LET dataset aims to address these issues by providing a standardized, high-quality data resource that enhances data circulation and value in the humanoid robotics sector [3] Group 3: Ecosystem Development - The LET dataset will be continuously maintained and updated under the Open Atom Open Source Foundation, contributing to a systematic resource for real-world data in the industry [4] - The integration of the LET dataset into the OpenLoong community will facilitate deeper research in task modeling, skill learning, and strategy validation, while providing high-quality samples for performance verification [4]

LET数据集在场景覆盖上跨越工业、商业零售和日常生活三大领域，包含汽车工厂、快消场景、酒店服 务、3C工厂、生活服务和物流共六类真实生产与服务环境，并在此基础上构建了31项任务和117种原子 技能，形成了结构清晰的任务体系。这些任务目标清晰、流程有约束，适用于训练机器人在多场景、多 步骤、多目标条件下进行学习与推理。 与此同时，数据集同步记录头部与双腕视觉流，并提供RGB、深度、关节状态和末端执行器状态等多 模态信息；通过组帧技术将传感器时间误差压到10毫秒以内，再以视觉伺服闭环机制对齐到统一抽象空 间，使数据一致性超过90%。 此次被捐赠的LET数据集由乐聚智能与合作团队联合构建，是国内少有的面向真实作业场景的全尺寸人 形机器人真机数据集，它基于乐聚夸父系列全尺寸人形机器人进行统一采集，覆盖真实作业链路，规模 超过6万分钟。 证券时报记者从国家地方共建人形机器人创新中心(简称"国地中心")获悉，近日，乐聚智能正式将其自 主构建的LET数据集捐赠至OpenLoong开源社区，为社区补充了一套覆盖真实作业场景的重要数据资 源。该数据集规模大、结构规范、场景丰富，具备突出的工程化与研究价值，将为国内人形机器人研发 ...

Core Viewpoint - The article emphasizes the importance of high-quality, multi-modal, and structured data in advancing humanoid robot technology and its applications, highlighting the donation of the LET dataset to the OpenLoong open-source community as a significant step towards building a unified data infrastructure in the industry [1][11]. Group 1: LET Dataset Overview - The LET dataset, constructed by Leju Intelligent and its partners, is one of the few full-size humanoid robot datasets in China, covering real operational scenarios with over 60,000 minutes of data [2]. - The dataset encompasses diverse task scenarios across three main fields: industrial, commercial retail, and daily life, including six categories such as automotive factories and logistics, with a clear task system comprising 31 tasks and 117 atomic skills [4]. Group 2: Data Collection and Technology Innovation - The dataset records both head and dual-wrist visual streams, providing multi-modal information such as RGB, depth, joint states, and end-effector states, achieving data consistency over 90% through advanced frame grouping technology [5][6]. - Complex tasks are broken down into clearly defined atomic action steps, accompanied by semantic labels, facilitating model understanding of task structures and action logic, thus enhancing behavior understanding and skill learning [7]. Group 3: Building a Trustworthy Data Standard System - The humanoid robot industry faces challenges such as fragmented data sources and inconsistent formats, necessitating a systematic data standard to enhance data quality and model capabilities [12]. - The National and Local Joint Innovation Center for Humanoid Robots has established a comprehensive standard system covering data collection, processing, quality review, and version management, ensuring data quality and usability [14]. Group 4: OpenLoong Community and Future Prospects - The donation of the LET dataset to the OpenLoong community enriches the repository of real-world data, promoting deeper research in task modeling, skill learning, and strategy validation [11][24]. - OpenLoong aims to create a shared data framework to standardize data organization and reuse, with the LET dataset serving as a representative training sample for the industry [20][26].

Group 1 - The core viewpoint of the article highlights the rapid development and innovation in the robotics industry, particularly focusing on humanoid robots and intelligent systems [5][11][14]. - The launch of the LinkSoul platform by Zhiyuan Robotics allows users to customize robot features, including voice and behavior, enhancing user interaction [2]. - The Anhui provincial government has introduced an action plan aiming to cultivate over 10 leading domestic enterprises and achieve a total industry revenue of 100 billion yuan by 2027 [5]. Group 2 - GAC Group plans to initiate large-scale production of its embodied intelligent robots by 2027, with a target of exceeding 1 billion yuan in industry chain output by 2030 [8]. - The LET dataset, developed by the National Local Joint Innovation Center for Humanoid Robots and Leju Intelligent, aims to address the lack of real machine data in the field of embodied intelligence [11]. - Dongfeng Motor has unveiled several humanoid robots, including "Xiao Dong" and "Worker No. 2," which are designed for tasks such as customer reception and factory operations [14].

在全球人工智能技术从虚拟感知迈向物理交互的临界点，具身智能的浪潮正以前所未有的势头重塑科技疆界。 然而，这条通往通用机器人的道路上，始终横亘着一道难以逾越的鸿沟 ， 那就是 高质量、大规模、标准化 的真实世界操作数据极度稀缺。数据的匮乏，如同燃料短缺，严重制约着人形机器人模型能力的跃迁与规模化 落地。 近日，全球首个面向真实作业场景的全尺寸人形机器人真机数据集 ——"LET数据集"正式发布， 并开源首批 60,000+分钟数据 。 这不仅是国内开源规模最大的全尺寸人形机器人真机数据集，更以其超过六万分钟的真 机实采数据、多模态融合架构与全场景覆盖能力，精准命中了行业最核心的痛点。它的出现，并非简单增加了 一个数据来源，而是为整个具身智能领域注入了关键的"数据新燃料"，有望驱动一场从"机械执行"到"自主理 解与推理"的范式革命。 ▍ 人形机器人数据集为何稀缺？ 人形机器人数据为何如此珍贵且稀缺？其背后是技术与成本的双重高墙。 在人工智能领域广为人知的 "Scaling Law （ 缩放定律 ） "指出，模型性能随着数据量、模型规模和计算力的增加而显著提升。这一法 则同样适用于正处于爆发前夜的人形机器人与具身智能 ...