阿里终于入局具身智能了。 近日 ，具身智能公司「 自变量机器人 」完成 十 亿元 A + 轮融资，本轮由 阿里云、国科投资 领投，国开 金融、红杉中国、渶策资本跟投。老股东美团战投超额跟投，联想之星、君联资本持续追投。本轮融资将用于 自变量全自研通用具身智能基础模型的持续训练和硬件产品的研发迭代。 自 2023 年初成立以来，自变量的融资节奏堪称迅猛 。 短短 两 年多时间已完成 8 轮融资，累计金额超 20 亿元，投资方覆盖顶级 VC 、产业资本与国资基金。 尤其在 202 5 年，公司更是以每轮数亿元的规模连续 完成四轮融资 。 这样的融资频率和规模，不仅在初创企业中罕见，更印证了资本市场对其技术路线和商业化 潜力的高度认可。 | | | | | 阿里云 | | 阿里云 | | | --- | --- | --- | --- | --- | --- | --- | --- | | | | | | 国科投资 | 一笔交易 阿里领 投,自变量机器人完 | 国科投资 | | | 1 | 2025-09-08 | A++ | 近10亿人民币 | 国开剑 | | 但有意識 | 成近10亿元A+轮融 | | | | ...

8 月 28 日，埃斯顿发布 2025 年半年报。报告显示， 公司上半年营业收入为 25.49 亿元，同比增长 17.50% ；归母净利润为 668.23 万元，同比增长 109.10% 。 分地区来看， 2025 年上半年国内业务收 入 18.00 亿元，同比增长 25.31% ；海外业务收入 7.49 亿元，同比增长 2.18% 。 值得注意的是， 据 MIR 睿工业数据统计， 2025 年上半年埃斯顿工业机器人出货量首次超越外资品牌，成 为首家登顶中国工业机器人市场的国产机器人品牌，市场份额进一步提升 ，其市场在汽车、电子、锂电等应 用领域实现快速增长。 从业务类型来看，埃斯顿工业机器人及智能制造系统业务收入达 20.92 亿元，同比增长 26.54% ， 这一增 长幅度远超公司整体营业收入 17.50% 的增幅，表明机器人以及集成板块 依然 是公司业绩增长的重要驱动 力 ， 2025 年工业机器人 市场需求回暖及国产替代进程加速，大环境好转下，使得埃斯顿工业机器人出货 量持续增长，并在上半年取得历史性突破。 截至 2025 年上半年，埃斯顿凭借连续两个季度位列中国机器人市场第一的成绩， 市场份额达 1 ...

2025年 9 月 5 日， 机器人大讲堂获悉， 行业 首个开放架构超大规模 多模态 触觉数据引擎与认知训练中 心 （ 以下简称 "触觉训练中心" ） 在湖北成立，这是我国机器人领域又一重磅布局 ， 该 训练中心由 湖北 人形机器人创新中心、武汉华威科智能技术有限公司 、 湖北人形机器人产业联盟 三方合作 共建 ， 有望 填 补国内真人触觉对齐数据空白，破解行业长期存在的触觉数据碎片化 、 共建共享协同难题，为人形机器人产 业构建核心感知基座。 ▍ 触觉为何成为人形机器人发展重心？ 业界对于人形机器人触觉正呈现越来越高的关注度。 近期， X 平台有网友曝出特斯拉疑似 Optimus 3 人形 机器人的原型， 此前就展示过如 "接网球"类似依赖实时触觉反馈的动作 ， 网友猜测该款机器人 或 多处采 用电子皮肤，构建触觉体系。 人形机器人触觉系统之所以得到了全球重点企业的重视，一方面由于人类人体 的触觉感知功能多样且复杂 ， 触觉 一直是 人类感知系统的核心组成部分 存在 ， 能够为 人形机器人的 交互 感知 和自动化系统提供重要 的信息输入手段 ， 更加准确地感知物体的位置和表面属性。 另一方面， 人形机器人 ...

2025无锡国际人工智能创新应用大赛于8月25日正式开赛，大赛面向全球开放算法赛道和具身智能创新应用 赛道双赛道，召唤广大算法开发者、创新团队、科研院所和企业共同参与这场具身AI大赛，实现人工智能技术 创新与应用。 具身 前沿 赛题 ， AI 创新挑战 本次大赛聚焦具身智能领域，算法赛道参赛者将使用极市平台和 DISCOVERSE 具身仿真平台进行 算法开发 角逐。 具身智能创新应用赛道面向具身智能创新应用企业、具身生态链企业、智能终端企业、具备创新思维的创业团 队、科研院所团队和个人，基于具身智能进行创新和应用开发，提出并实现具有创新性和实用价值的解决方 案。 现在报名，与全球顶尖人才 和具身智能生态企业 同台竞技，共同开创具身智能的崭新时代！ 赛事官网： https://cvmart.net/cv_landing/list/wuxi2025 赛题概况 【 算法赛道 】 机器人原料识别赛题： 机器人原料识别算法致力于精准识别生产线上或特定场景中的原料，助力提高资源利用率与生产效率。通过视 觉系统获取物料图像，然后运用图像处理与分析技术，辨别出哪些是原料，并结合机器学习模型，对原料的类 型、形状、位置等特征进 ...

2023 年，瑞金医院 敏锐觉察到锟铻 ® 国产手术机器人技术的潜力，经一系列学习验证， 5 月 11 日 由 骨 科何川教授团队 成功采用锟铻 ® 国产机器人， 开展了首例骨科机器人辅助手术 ，迈出了 瑞金医院 骨科手 术革新的一大步 。 12 月 8 日 ， 何川教授团队成功实施了 100 例国产机器人辅助下人工关节置换术，凸显 了 在关节外科领域的领先地位。 9 月 5 日，上海交通大学医学院附属瑞金医院 骨科何川教授团队 宣布 成功完成 300 台骨科机器人手术，标 志着瑞金医院骨科在智能手术领域实现关键突破，这一创举不仅是瑞金医院的辉煌成就， 更是 代表着国产手 术机器人系统在临床中的应用已从 "探索期"走向"成熟期"。 值得一提的是， 300 台手术 涵盖了骨科领域多种重要术式。其中，初次全膝关节置换术占比约 70% ，初次 全髋关节置换术占比 20% 以上，剩余为初次单髁膝关节置换术及探索性的翻修手术 ，证明了锟铻 ® 全骨科 手术机器人 的多适应症优秀特性。 ▍ 从 0 台到 300 台，骨科手术机器人成新方向 在中国的骨科领域，瑞金医院占据着举足轻重的地位。 早在上世纪 他们在国内较早涉足髋 ...

Core Viewpoint - Midea Group has opened its first intelligent factory in Jingzhou, Hubei Province, which is recognized as the world's first multi-scenario intelligent factory, showcasing advancements in autonomous industrial ecosystems and surpassing traditional digital factories [1][2]. Group 1: Factory Overview - The intelligent factory has been awarded the title of "World's First Multi-Scenario Intelligent Factory" by the World Record Certification Agency (WRCA), highlighting its efficiency, flexibility, and resilience, and marking China's leadership in smart manufacturing technology [2]. - The factory employs various intelligent products, including the Miro humanoid robot, AMR robots, and AI glasses, all coordinated by a "factory brain" that manages operations and decision-making [2][3]. Group 2: Efficiency and Performance - Since implementing AI solutions, the factory has seen an average efficiency increase of over 80% in core processes, with a 100% error-proof rate in key operations, and response times reduced from hours to seconds [3]. - The factory's intelligent systems allow for seamless operations, with Miro capable of transporting components and performing safety inspections, while AMR robots navigate production areas autonomously [5][8]. Group 3: Future Prospects - Midea plans to replicate the intelligent factory model across its more than 100 global factories, although the full-scale adoption of humanoid robots and AI technologies will take time due to ongoing optimization needs [12]. - The company reported a 14.3% year-on-year profit increase to 38.5 billion RMB (5.4 billion USD) last year, with continued growth in 2023, indicating the financial benefits of its AI applications [12][13].

Core Viewpoint - The debut of the "Machine Wolf" during the military parade marks a significant milestone in China's military modernization and the development of bionic intelligent robots, showcasing the country's advancements in intelligent warfare capabilities [1]. Group 1: Machine Wolf's Capabilities - The "Machine Wolf" exhibits superior combat capabilities compared to the "Machine Dog," making it more effective in various operational scenarios [5]. - In urban warfare, the "Machine Wolf" can navigate narrow spaces that are difficult for humans, allowing it to perform tasks in high-risk areas and work effectively alongside human soldiers [7]. - A complete "Machine Wolf" combat group consists of a control vehicle and several four-legged robots, enabling seamless communication and dynamic collaboration among "human-vehicle-wolf" [7]. Group 2: Technological Features - The Q20A platform, which supports the "Machine Wolf," features an open architecture with standard software and hardware development interfaces, ensuring high mobility and adaptability in complex environments [10]. - The platform is equipped with a 360° visual perception system and high-precision laser radar modules, providing exceptional sensing capabilities [10]. Group 3: Application Scenarios - The Q20A robot is designed for various military and public safety applications, including reconnaissance, combat support, and emergency response [8][18]. - The technology has been tested in extreme environments, demonstrating its reliability and effectiveness in real-world scenarios, such as high-altitude operations and urban security [13][22]. - The Q20A has been integrated into public safety solutions across 47 cities, providing a comprehensive approach to security and emergency management [15]. Group 4: Industry Impact - The current phase represents a critical moment for the bionic intelligent robot industry, focusing on scaling production and real-world applications [22]. - The Q20A's modular design lowers the barriers and costs for industry applications, facilitating the widespread adoption of bionic robots across various sectors [24]. - The advancements in the Q20A platform signify a transformation from traditional equipment to new forms of combat power, reshaping future battlefield dynamics [24].

中航迈特专注金属 3D打印设备与材料技术创新与产品服务，聚焦具身智能机器人行业，可为该行业提供从原 型开发到批量生产的金属3D打印装备及金属粉末材料支持。 中航迈特将在 2025年TCT深圳展3C12展位 上，为机器人行业创新带来金属3D打印智造解决方案。 展会上，中航迈特将携 MT170H、MT280及MT400M金属3D打印机 等闪耀现场，并将于现场演示机器人身 体部件的一体化成形打印。系列设备满足多样化机器人制造需求，支持复杂结构一体化、产线化批量生产，为 机器人制造企业提供可靠的生产解决方案，让具身体制造更自由、更高效。 小指如章鱼般柔软蜿蜒，指尖还自带 "透视眼"，整只"手"遇热能变色提醒，单根"手指"成本不到1块钱。 在 2025世界人工智能大会（WAIC）上， 复旦大学智能机器人与先进制造创新学院智能机器人研究院甘中学 教授团队自主研发的多源仿生 3D打印柔性灵巧手首次亮相 。 这款灵巧手打破传统设计，融合仿生结构、 3D打印柔性材料与传感控制系统，不仅能灵活抓握、夹取、勾 挑，还具备温度变色、视觉识别等功能，巧妙应对复杂环境，同时大幅降低制作成本，诠释机器人与人类共融 共生的美好未来。 深圳， ...

Core Insights - The article discusses the challenges in robotic manipulation and introduces a new system called HIL-SERL that significantly improves the efficiency and effectiveness of robotic training in real-world scenarios [1][2]. Traditional Methods Challenges - Traditional robotic control methods require extensive engineering design or imitation learning, which often lack adaptability and struggle in new environments [1]. - These methods fail to achieve human-level proficiency and speed, leading to inefficiencies in real-world applications [1]. HIL-SERL System Overview - The HIL-SERL system developed by a research team at UC Berkeley allows robots to learn complex tasks with only 1 to 2.5 hours of real-world training, achieving near-perfect success rates and surpassing human execution speeds [2][3]. - The system combines human guidance with autonomous exploration, creating an efficient and safe learning loop [3]. System Architecture - HIL-SERL consists of three core components: an executor process, a learner process, and a replay buffer integrated within the learner [4]. - It employs off-policy reinforcement learning techniques to optimize behavior strategies by leveraging historical data, allowing robots to learn from human demonstrations and assess the contribution of different actions towards achieving goals [4]. Performance in Multi-Task Scenarios - The system was tested on challenging tasks such as precision assembly, dual-arm coordination, and dynamic manipulation, demonstrating its versatility [5][8]. - In precision assembly tasks, robots achieved sub-millimeter accuracy, while in dual-arm coordination tasks, they effectively managed complex operations requiring synchronized movements [8]. Results and Adaptability - After 1 to 2.5 hours of training, robots achieved nearly 100% success rates and executed tasks 1.8 times faster than traditional imitation learning methods, which had an average success rate of 49.7% [9]. - The robots exhibited remarkable adaptability, successfully adjusting to unexpected situations, such as misalignments or disturbances, showcasing their ability to learn from real-time feedback [12]. Learning Mechanism - HIL-SERL's adaptability stems from its ability to evolve different control strategies based on task requirements, allowing for real-time adjustments and corrections [13][16]. - For high-precision tasks, the system employs a closed-loop response strategy, while for dynamic tasks, it utilizes an open-loop predictive strategy, demonstrating a high level of confidence in executing planned actions [13]. Conclusion - The research highlights the potential of HIL-SERL to overcome traditional reinforcement learning limitations, enabling efficient learning of complex skills in real-world environments [14]. - This advancement opens new avenues for industrial applications, particularly in flexible manufacturing sectors requiring small-batch production [14].

Core Viewpoint - The article highlights the cybersecurity vulnerabilities faced by robotic companies, particularly in the context of Pudu Robotics, as they expand globally and integrate into sensitive environments like restaurants and hospitals [1][11]. Summary by Sections Security Vulnerabilities - Bobdahacker exposed significant security flaws in McDonald's ordering system and Pudu's AI chatbot, revealing that simple passwords like "123456" could be exploited to gain unauthorized access [1][3]. - The vulnerabilities allow attackers to control food delivery robots, redirect orders, and disrupt restaurant operations through DDoS attacks [3][5]. Attack Capabilities - Attackers can view call history for any robot and retrieve up to 20,000 store IDs in a single request [5]. - They can initiate, cancel, or reschedule tasks for any robot globally [6]. - Modifications to robot settings, including nicknames and operational behaviors, are also possible [7]. Broader Implications - The security risks extend beyond restaurant chaos, potentially affecting hospitals relying on robots for medication delivery, leading to treatment delays or misdelivery [8]. - Pudu Robotics, the largest commercial service robot manufacturer, faced scrutiny after failing to respond promptly to vulnerability reports, only taking action after warnings from major clients [9][10]. Industry Challenges - The article emphasizes that many robotic companies lack basic security measures, such as dedicated security contacts and authenticated API controls, often only responding to threats when reputational damage is imminent [12]. - As automation plays a larger role in critical operations, the need for robust security capabilities that match technological innovations becomes increasingly urgent [12].