Core Viewpoint - The article discusses the emerging trend of humanoid robots entering households, highlighting both the excitement and skepticism surrounding this development. It emphasizes the need for a more user-centered approach in the design and deployment of household robots, as current trends may not align with user preferences and safety concerns [1][12]. Group 1: Expert Insights - Maya Cakmak, a professor at the University of Washington, expresses her concerns about the practicality of humanoid robots in homes, suggesting that non-humanoid robots may be more effective and user-friendly [3][4]. - Cakmak's research indicates a significant preference among users for specialized robots over humanoid ones, citing safety, privacy, and space concerns as primary reasons [6][8]. Group 2: User Preferences - A study involving 76 participants revealed that most preferred specialized robots for household tasks, associating humanoid robots with safety risks and privacy issues [6][7]. - Users expressed that specialized robots are perceived as safer, more private, and less intrusive compared to humanoid robots, which are often seen as bulky and unnecessary [6][11]. Group 3: Practicality vs. Design - The article highlights a critical contradiction in the industry: while companies pursue humanoid robots for their versatility, users may not require human-like features for effective assistance in household tasks [9][11]. - Cakmak argues that simpler, non-humanoid designs could fulfill most household needs without the complications associated with humanoid robots [11][12]. Group 4: Future Directions - The article concludes that the industry should focus more on user insights and practical applications rather than merely technological advancements. It suggests that companies should share user research data to better align product development with actual user needs [12][14].

Core Viewpoint - The FCloud OmniBot Empowerment Salon focused on the development of embodied intelligence technologies, emphasizing the importance of physical simulation and data synthesis for scaling applications in the robotics industry [1][3][20]. Group 1: Industry Development Opportunities - The event gathered experts from academia, research institutions, and industry to discuss new opportunities for industrial development [3][5]. - Zhangjiang Science City has over 1,000 AI companies, with more than 90 in the field of embodied intelligence, forming a complete industrial chain from core components to complete machine development [5][20]. Group 2: FCloud OmniBot Platform - FCloud has established a 2,000-card computing center in Zhangjiang to support local enterprises, with plans for further expansion [7][9]. - The OmniBot platform addresses three main challenges in embodied intelligence development: simulation environment setup, synthetic data generation, and computing power requirements [9][20]. - OmniBot integrates NVIDIA Isaac Sim and Isaac Lab for high-performance simulation capabilities, allowing developers to access simulation software via cloud desktops without complex local setups [11][20]. Group 3: Technical Innovations - The platform can generate 100 synthetic data points from a single real-world data point, significantly enhancing data collection efficiency [11][20]. - OmniBot supports cloud training and deployment of mainstream models, including specialized models for embodied intelligence [12][20]. - The cloud-edge collaboration model allows developers to train models in the cloud and deploy them on robots, reducing development costs and barriers [12][20]. Group 4: Academic and Technical Sharing - The salon featured discussions on the data gap in the robotics field, highlighting that training data for robots is 6,500 times less than that for large language models [13][20]. - Research from Shanghai Jiao Tong University introduced a novel instruction expression method that improves efficiency and generalization capabilities [15][20]. Group 5: Open Ecosystem and Collaboration - FCloud OmniBot emphasizes ecosystem development, welcoming partnerships from various stakeholders, including robot manufacturers and algorithm developers [18][20]. - The platform operates on a SaaS model, providing flexible access and special policies for students and individual developers to encourage participation [18][20]. Group 6: Future Trends and Prospects - The trend towards simulation-first development is becoming mainstream, with physical simulation seen as key to addressing data scarcity and reducing development costs [20]. - The integration of cloud-edge collaboration is essential for meeting the increasing complexity of robotic tasks [20]. - The continuous decline in computing costs and improvements in simulation technology are expected to lead to large-scale applications of embodied intelligence within the next 3-5 years [20][21].

Core Insights - The article discusses the advancements in embodied intelligence, particularly focusing on the new training paradigm introduced by Zhujidongli with their LimX DreamActor, which utilizes a multi-data approach for training robots [1][3][25]. Group 1: LimX DreamActor Overview - LimX DreamActor integrates video data, simulation data, and real machine data to enhance robot training efficiency and performance [3][17]. - The training process consists of four steps: data collection using consumer-grade devices, 3D reconstruction with physical parameters, extensive training in simulated environments, and fine-tuning on real machines [7][9][15]. Group 2: Data Utilization Strategy - The multi-data strategy addresses the limitations of each data type: real machine data is expensive, simulation data lacks realism, and video data is challenging to apply due to the absence of physical properties [3][17]. - The approach emphasizes data efficiency, aiming to achieve better performance at lower costs by leveraging diverse data sources [3][16]. Group 3: Technical Innovations - DreamActor employs advanced real machine reinforcement learning (RL) techniques, which significantly enhance learning efficiency and the ability to generalize from simulation to real-world applications [16][18]. - The integration of Real2Sim2Real strategies allows for a more reliable deployment of robots, reducing risks and shortening development cycles [18][20]. Group 4: Historical Context and Evolution - LimX DreamActor is an evolution of previous efforts by Zhujidongli, such as LimX VGM, which focused solely on video data for training robots without real machine samples [21][23]. - The transition from VGM to DreamActor reflects a deeper understanding of data application and the pursuit of optimal data-performance ROI [21][23]. Group 5: Industry Implications - The advancements in the multi-data approach are expected to lower the barriers for participation in embodied intelligence development, enabling more teams to engage in this field [25]. - The article suggests that achieving a balance between efficiency and stability in robot training is crucial for the large-scale application of embodied intelligence [25].

Core Viewpoint - Continuous robots exhibit great potential in fields such as robotic surgery and narrow space detection, but precise control remains a significant challenge. Recent breakthroughs by research teams from City University of Hong Kong and Hefei University of Technology have applied Kalman filtering technology to enhance the online control precision of these robots [1][2]. Group 1: Continuous Robots and Control Challenges - Continuous robots possess infinite degrees of freedom and adaptability, making them difficult to control accurately due to their deformable nature, akin to controlling a cooked noodle [4]. - Traditional rigid-link robots have simpler control mechanisms, while continuous robots face challenges from large deformations, friction effects, and inherent non-linear characteristics [4][5]. - The research team designed a lightweight robot with a complex internal structure, consisting of three flexible segments, each with five spacer disks and one drive disk, weighing only 8.4 grams [5]. Group 2: Control Methodology - The team utilized a piecewise constant curvature (PCC) model for initial control, which, while computationally efficient, resulted in position errors exceeding 1.6 mm and angle errors over 1.4 degrees, unacceptable for high-precision applications [7]. - Instead of developing a more complex model, the team innovatively employed the Kalman filter to allow the robot to self-correct during motion, estimating and compensating for errors in real-time [8][9]. - The control system operates at a frequency of 20 Hz, integrating steps such as obtaining end pose, calculating model Jacobians, estimating and compensating for Jacobian errors, and generating control commands [11]. Group 3: Experimental Validation - The research team conducted three trajectory tracking experiments and two disturbance resistance tests, demonstrating the effectiveness of the new method [12]. - In the first experiment, the root mean square error (RMSE) in the x-direction improved from 1.6 mm to 1.1 mm, and in the y-direction from 2.3 mm to 2.1 mm, showcasing significant enhancements in tracking precision [12][14]. - The second experiment focused on attitude control, achieving a reduction in RMSE from 2.1 degrees to 1.5 degrees, while maintaining position accuracy [14]. - The robustness of the method was further validated through disturbance tests, where the robot maintained performance even under significant load changes [15]. Group 4: Innovation and Future Prospects - The research combines model-driven and data-driven approaches, leveraging the strengths of both to enhance control precision while maintaining computational efficiency [17]. - The method's advantages include no need for offline data collection, high computational efficiency, and robustness against external disturbances, indicating strong potential for practical applications [17]. - Future research directions include incorporating dynamic effects and expanding to three-dimensional motion to improve estimation accuracy and applicability [17].

Core Viewpoint - The 2025 Science & Technology Innovator Conference focuses on "Embodied Intelligence as a New Engine for Industrial Transformation," emphasizing the commercialization of hard technology and addressing the challenges of transitioning from technology to product [3][5]. Event Overview - The conference is organized by the Zhiyou Yari Innovation Hub and will take place on September 5, 2025, at the Wanli Hotel in Beijing's Zhongguancun Dongsheng Science Park [6]. - The event will feature various segments including award ceremonies, report releases, keynote speeches, peak dialogues, roundtable forums, and thematic presentations, covering the entire value chain of embodied intelligence [5][6]. Key Themes and Services - The conference aims to create a "three libraries and four chains" service system, which includes an industry library for precise demand matching, a project library for selecting quality technologies, and a talent library for gathering innovative forces [3]. - The focus will be on providing a full-chain service model that includes demand-driven technology matching, capital support, and real-world scenario validation for cutting-edge technologies like embodied intelligence [3]. Notable Speakers and Topics - Keynote speakers include Paolo Dario, Huang Tiejun, and Wang Tiancao, who will discuss topics ranging from robotics' role in health to the evolution of embodied intelligence technology [7][8]. - The conference will also feature discussions on the commercialization challenges and breakthroughs in embodied intelligence, with insights from industry leaders and researchers [8][9]. Industry Impact - The conference is expected to attract top figures in the embodied intelligence field, facilitating knowledge exchange and strategic management insights from macro trends to micro case studies [5][8]. - It aims to address the "last mile" challenge in technology-to-product transition, providing a platform for real-world application and scaling of advanced technologies [3][5].

Group 1: Humanoid Robots - Ubtech has secured a procurement contract worth 250 million yuan for humanoid robot products and solutions, primarily featuring the Walker S2 model, with delivery planned for this year [1][3] - The total contract value for Ubtech's Walker series humanoid robots has reached nearly 400 million yuan, including approximately 50 million yuan in delivered orders [3] - Ubtech has established collaborations with companies like BYD, Foxconn, and SF Express in sectors such as new energy vehicle manufacturing, 3C manufacturing, and smart logistics [3] Group 2: Smart Elderly Care Robots - A joint expert group from the Ministry of Industry and Information Technology and the Ministry of Civil Affairs has selected 32 pilot projects for smart elderly care robots, indicating a shift towards large-scale growth in this industry [4][6] - The selected projects include intelligent emotional companionship robots, AI-assisted exoskeletons, and smart nursing collaborative elderly care robots [6] Group 3: AI Startups - French AI startup Mistral AI is completing a funding round of 2 billion euros, with an expected valuation of 14 billion dollars, making it one of the most valuable tech startups in Europe [7][9] - Mistral AI focuses on open-source language models and AI chatbots tailored for the European market, with its valuation previously at 5.8 billion euros [9] - Investment in European AI startups has seen a significant increase, with a 55% year-on-year growth in the first quarter of 2025 [9] Group 4: Robotics Research - A new multi-robot motion planning method based on Graph Neural Networks (GNN) and reinforcement learning has been published in Science Robotics, developed by Google DeepMind Robotics, Intrinsic, and University College London [13][15] - This method allows for the automatic generation of collision-free trajectories by mapping robots, tasks, and obstacles as graph nodes and edges, reducing the need for manual programming [15] Group 5: Investment in Robotics - Shoucheng Holdings has made a multi-million yuan additional investment in Songyan Power to accelerate humanoid robot research and development [10][12] - The investment aims to enhance R&D efforts and product iterations, solidifying Songyan Power's leading position in humanoid robots and bionic facial technology [12]

1、 中国人形机器人亮相上合组织峰会 近日，规模空前的上合组织峰会在天津举行，20多位外国领导人及10位国际组织负责人齐聚。天津梅江国 际会展中心二期的新闻中心科技感爆棚。一进门，一款高仿真人形智能机器人格外吸睛，其短发、红唇、 金属色躯干，能以中、俄、英三语互动，提供信息咨询、场地指引等全方位服务。此外，送物机器人穿梭 其间，8K技术结合创新交互带来独特体验，制作冰淇淋、写书法的机器人也各展身手，自动充电机器 人、"圣·宝莲尊"全自主无人机巡检系统等纷纷亮相。 2、 宇树科技预计第四季度申请IPO 据彭博引述消息人士报道，阿里巴巴支持的机器人公司Quicktron（快仓智能）已秘密申请香港IPO，最快 或于明年上市，预期募资至少1亿美元。据悉，快仓智能成立于2014年，是全球第二大智能仓储机器人系 统供货商，专注为客户提供以移动机械人集群及智能操作系统为核心的解决方案。历经数轮融资，其投资 者阵容强大，涵盖百世快递、菜鸟、软银中国资本等众多机构。值得一提的是，南山控股旗下宝湾物流为 快仓智能子公司提供3.5亿元银行授信担保，双方资本绑定紧密，共同打造的上海快仓宝湾产业园是合作 标志性项目。此次快仓智能申请香 ...

为了深入推动智能机器人全产业链区域协同和融合发展，支持新技术新产品在重点终端市场应用，工业和 信息化部产业发展促进中心湾区制造业创新成果产业化服务中心（简称"湾区中心"）拟于9月18日 重点围 绕智能机器人各细分方向，展开 全产业链优质项目供需与政策交流座谈会 。座谈会主要围绕政策解读、 应用落地及产融结合等方面，全方位介绍广州市黄埔区在智能机器人领域的政策体系，通过"机器人+"应 用示范带动效应，推动机器人产业链需求侧与供给侧的深度协同。 重点诚邀具备产业化落地意愿及诉求 的智能机器人产业领域优质创新成果项目/企业开展交流研讨座谈。 7月3日， 广州开发区、黄埔区在PCI未来社区举办 "人工智能+机器人"系列产业政策新闻发布会 ，重磅推出 "人工智能+机器人" 系列政策，即支持人工智能、具身智能、集成电路、超高清视频与新型显示产业高质量 发展若干政策措施， 形成了 4大政策近40条措施 ，最高给予 1 000万 元扶持 ，以政策创新激活产业动能。 据悉，这是黄埔区首次将政策新闻发布会开进企业园区，融合了政策宣讲、产品展示，第一时间将新鲜出炉的 政策送上门，现场吸引媒体朋友、相关企业、投资机构、金融机构、创业 ...

9月 施普林格·自然 特别推出 " AI慢思考：大模型复杂推理技术在线研讨会" 。本次活动特别邀请到 中国人 民大学高瓴人工智能学院赵鑫教授做客直播间，分享其团队围绕 DeepSeek-R1 等代表性模型在"慢思考"技 术方面的最新研究进展 。 同时，还邀请到 施普林格·自然计算机科学图书编辑总监常兰兰博士加入讨论，介 绍施普林格·自然计算机科学和2025年全新人工智能图书资源以及学术图书出版分享，并精选推荐相关领域的 图书 。 期待共同探索 AI 推理能力的新边界！ 随着人工智能技术的不断演进，大语言模型正逐步从简单的"快思考"模式，转向更注重推理深度与逻辑连贯性 的"慢思考"范式。以 DeepSeek-R1 为代表的新型大模型，通过引入强化学习机制强化长思维链推理能力， 在复杂任务处理上展现出更强的理解力和决策能力。这类模型不仅在数学、编程等高阶任务中表现优异，还以 开源架构和成本优势推动了技术的普及与应用。当前，"慢思考"技术正成为推动大模型迈向更高智能水平的关 键路径，引领行业走向更具自动化与可靠性的未来。 研讨会日程 开场致辞 大模型慢思考技术探讨 赵鑫教授，中国人民大学高瓴人工智能学院 施普林格· ...

在工业制造的版图里，焊接是绕不开的关键一环，却常年被三大痛点牢牢困住：要靠大量人工 作业 、焊接质 量时好时坏、工人还得在高温、高辐射的恶劣环境里作业。 如今， 随着 机器人 自动化与人工智能技术的爆发式发展，焊接行业正迎来一场由技术创新驱动的大变革 ， 一家名为【 集萃智造 】的企业，正尝试 用一款会自己思考的焊接机器人，重新定义智能制造的边界。 ▍ 四次迭代，啃下卡脖子技术 据机器人大讲堂了解， 早在 2019 年，集萃智造就推出了第一代协作焊接机器人。此后五年里，他们没停下 升级的脚步，一路迭代四次，硬是造出了五款不同负载的机械臂，覆盖了从轻型到重型的多种焊接场景。 更关键的是，他们没走组装 厂 路线，而是 努力 把核心技术攥在了自己手里： 公司坚持核心零部件自主研 发， 硬件上 成功 攻克了直流无框电机、高精度双磁编码器等关键部件， 构建了其 机器人稳、准、快的基 础；软件上 ， 该公司还搞定了 机器人动力学建模、智能焊接工艺库，甚至搭起了跨平台操作系统 ， 相当于 给机器人装了最强大脑。 据悉，如今 现在，集萃智造的便携式协作焊接机器人已经扎根重工业：港口机械的大型构件、水利工程的金 属结构、钢结构 ...