Core Viewpoint - The article discusses the rapid development of soft wearable robots, also known as exosuits, which are revolutionizing rehabilitation technology by providing more comfortable and effective assistance compared to traditional rigid exoskeletons [1][2]. Group 1: Transition from Rigid to Soft Rehabilitation Robots - Traditional rehabilitation robots often use rigid exoskeleton structures, which can be complex and heavy, increasing metabolic costs and limiting their use to clinical environments [2]. - Soft exosuits are designed to wrap around the user's body, working in synergy with the user's muscles and preserving their natural range of motion, thus enhancing comfort [2]. Group 2: Overview of Actuation Technologies - The review categorizes soft exosuit actuation technologies into four main types, each with unique characteristics and applications [4]. - Fluid-driven actuators are the most widely used, with 55 devices employing this technology, primarily pneumatic drives, which can mimic complex muscle movements but face portability challenges [5]. - Motor-cable drives, used in 59 devices, offer precise control and high bandwidth but may cause friction and skin damage [7]. - Shape memory alloys (SMA) provide quiet and efficient operation but have limitations in response speed and control [8]. - Dielectric elastomer actuators (DEA) represent cutting-edge technology with impressive performance metrics but pose safety concerns due to high operating voltages [9]. Group 3: Joint Assistance Solutions - The review systematically analyzes assistive devices for various joints, focusing on both single-joint and multi-joint collaborative assistance [12]. - Upper limb assistance primarily utilizes pneumatic drives, with innovative designs for shoulder and elbow support [13][15]. - Hand rehabilitation devices are in high demand, with over 40 types analyzed, showcasing various technological approaches [18]. - Lower limb assistance systems require greater force, with specific designs for hip, knee, and ankle joints demonstrating significant reductions in metabolic costs during walking [24][27]. Group 4: Trends in Hybrid Designs - A notable trend is the integration of hybrid designs, combining soft and rigid components to enhance mechanical efficiency while maintaining comfort [30]. - Examples include hand devices that guide tendons with reinforced structures and semi-rigid knee devices that improve force transmission efficiency [30]. Group 5: Commercialization Progress and Challenges - The review tracks the commercialization of soft wearable robots, highlighting products like ReWalk Robotics' ReStore and Carbonhand, which assist stroke patients in improving mobility and grip [31]. - Challenges remain in ensuring comfort while providing sufficient assistance, as well as addressing the limitations of various drive systems [33]. Group 6: Future Outlook - The article emphasizes the need for developing cable-free devices that can assist multiple joints while being intuitive and comfortable for users [34]. - This requires interdisciplinary collaboration in materials science, mechanical engineering, and control theory to create soft yet strong materials and intelligent control systems [34]. Group 7: Conclusion - Soft wearable robots represent the future of rehabilitation technology, with ongoing advancements expected to significantly improve the lives of millions with mobility impairments [36].

Core Viewpoint - The release of "Zhiyuan Lingqu OS," the first embodied intelligence operating system reference framework, marks a significant step towards standardization, scalability, and ecological development in the embodied intelligence industry [1][2]. Layered Open Source and Ecosystem Construction - "Zhiyuan Lingqu OS" promotes a "layered open source, co-construction and sharing" model, enhancing the existing high-performance middleware AimRT for stable and efficient distributed communication and hardware abstraction [2]. - The middle layer supports standardized connections for multi-modal perception, task planning, and action control, significantly lowering the algorithm deployment threshold for developers [2]. - The upper layer provides a comprehensive toolchain for simulation testing, data labeling, and model training, facilitating a technical closed loop from virtual simulation to physical deployment [2]. Accelerating the Popularization of Embodied Intelligent Robots - The introduction of "Lingqu OS" addresses issues such as fragmented system ecology and slow technological innovation in the robotics industry, serving as a foundation for building an industrial ecosystem [4]. - The open-source model is expected to attract global developers, fostering breakthroughs in challenges related to intelligent enhancement, group collaboration, and cloud-edge integration [4]. - The open-source plan for "Zhiyuan Lingqu OS" will begin in the fourth quarter of this year, aiming to witness transformative changes in the embodied intelligence field and propel the robotics industry to new heights [4].

Core Viewpoint - The trend of humanoid robot companies venturing into quadruped robots is emerging, indicating a shift in the robotics industry towards diversified product offerings and technological synergies [4][12]. Group 1: Product Launches - Zhiyuan Robotics launched its first quadruped robot, D1 Ultra, designed for special and industrial applications, featuring a maximum running speed of 3.7 m/s and the ability to jump up to 35 cm [2]. - MagicLab introduced its new wheeled quadruped robot, MagicDog-W, which boasts 17 degrees of freedom and can navigate complex terrains, with a starting price of 75,000 yuan [2]. Group 2: Technological Synergies - Quadruped and humanoid robots share a high degree of technological commonality, with approximately 60% of their components being similar, allowing for cost-effective research and development [5]. - The experience gained from developing quadruped robots can be directly applied to humanoid robots, facilitating faster product iterations and reducing development costs [6][9]. Group 3: Market Dynamics - The quadruped robot market is becoming increasingly competitive, with companies like Zhiyuan and MagicLab entering the space, which may lead to market homogenization [12]. - Zhiyuan holds a significant market share of nearly 70% in the quadruped robot sector, with a target production capacity of 5,000 units for 2025, indicating strong competitive positioning [6][13]. Group 4: Future Outlook - The global quadruped robot market is projected to exceed 8 billion yuan by 2030, with a compound annual growth rate (CAGR) of over 30%, positioning leading companies like Zhiyuan to capture more than 50% of the market [13]. - The integration of quadruped and humanoid robots into a dual product matrix is expected to provide stable cash flow and open up significant market opportunities [10].

7月26日，为期三天的 2025世界 人工智能大 会（WAIC 2025） 在上海顺利举行， 节卡机器人 作为马桥人 工智能创新试验区（展位号H3 616）代表企业之一，携 具身智能平台JAKA Lu m i 亮相。 当下， 具身智能已十分火爆，人们期待着机器人通过多模态感知、自主行动与决策和环境动态交互等多方面 能力的进化，实现复杂场景下的精准作业能力和强大的场景泛化能力，既要精准拿捏工业装配、质检环节的零 误差生产，又可化身新时代果农精准采摘鲜美果蔬，更要成为精密实验室的最佳助手。 然而，单一形态无法通吃所有场景。在具身智能的布局上， 节卡机器人提供的不是单一人形，而是一支能力 互补的 "具身智能特工队" 。 ▍ 双臂JAKA K1：身怀金刚钻，勇担瓷器活 1）高刚性高精度： 5kg负载，一体化关节设计，土0.05mm精准定位 2）EtherCAT通讯： 1ms通讯频率高速响应 3）末端内嵌六维力传感器： 0.1N分辨率，最小1N力控精度,导纳控制算法 4）高性能一拖二控制柜： 高性能控制器RoboHub，可支持64轴，外部拓展丰富 ▍ 轮式人形JAKA K1W：工厂多面手，骨骼之上塑血肉 实践是检验真理 ...

Core Viewpoint - The article discusses the innovative use of a drone system named Pei-Wo Drone developed by a research team from National Cheng Kung University in Taiwan, aimed at assisting elderly individuals in maintaining their exercise routines at home, addressing the global challenge of elderly inactivity [1][16]. Group 1: Technology and Design - The Pei-Wo Drone is a small, lightweight (30 grams) drone that utilizes advanced sensors for precise indoor positioning, including ToF sensors, optical flow sensors, and UWB positioning sensors [3][4]. - The system includes a drone, two wrist sensors, eight UWB positioning bases, and a control computer, allowing for real-time tracking of the user's hand positions [4][6]. - The drone operates by flying 1.2 meters in front of the user, providing a clear visual guide while minimizing psychological pressure [6][7]. Group 2: Testing and Results - A study involving 15 elderly participants demonstrated that the drone-guided exercise significantly improved accuracy, with vertical arm movements achieving an accuracy rate of 99.64% for the right arm and 92.35% for the left arm [8][9]. - The system's performance was evaluated through detailed motion analysis, identifying specific stages where participants struggled, which will inform future system enhancements [10]. Group 3: User Feedback and Acceptance - Participants expressed high satisfaction with the drone system, with many preferring it over traditional video guidance, citing ease of use and comfort during interaction [11][13]. - Suggestions for improvement included making the drone more visually appealing and incorporating video demonstrations for first-time users [14]. Group 4: Market Position and Future Directions - The Pei-Wo Drone offers distinct advantages over existing technologies like VR and AR, which can cause discomfort during prolonged use, by providing a real 3D guiding object that enhances the exercise experience [14]. - Future developments may include aesthetic enhancements, LED feedback, and adaptive algorithms to tailor the system to individual users' physical conditions and capabilities [15].

Core Viewpoint - The article highlights the emergence of humanoid robots, particularly the XMAN-F1 by Qianlang Intelligent, as a transformative force in service industries, moving beyond traditional manufacturing roles to enhance customer service experiences in various settings [25][27]. Group 1: Humanoid Robot Capabilities - The XMAN-F1 humanoid robot demonstrated its ability to perform tasks such as making popcorn and serving drinks, showcasing its potential in real-world service scenarios [2][4]. - The robot can adapt to complex environments, navigating obstacles and maintaining stability on uneven surfaces, which is crucial for operating in crowded spaces [10][12]. - XMAN-F1 employs a dual-layer model architecture that allows it to understand and execute service tasks effectively, transforming from basic actions to comprehensive service processes [14][15]. Group 2: Service Scene Integration - Qianlang Intelligent emphasizes the importance of integrating humanoid robots into various service scenarios, moving from flat to three-dimensional service models [8][10]. - The introduction of the ProS model allows robots to be trained for specific roles, enhancing their ability to perform tasks like bartending and medical logistics [12][14]. - The collaboration between humanoid robots and specialized robots, such as logistics robots, demonstrates the potential for creating efficient service ecosystems [18][20]. Group 3: Data Utilization and Continuous Improvement - Qianlang has deployed over 100,000 robots globally, generating more than 10TB of real-time data daily, which is used to train and improve the capabilities of the XMAN-F1 [22]. - The data-driven approach allows for continuous optimization of service processes, enhancing the robots' ability to adapt and perform in various environments [20][22]. - The integration of a feedback loop between real-world data and model training ensures that the robots remain aligned with service value and operational efficiency [22][29]. Group 4: Future Prospects - The article suggests that as humanoid robots like XMAN-F1 evolve, they will be capable of performing more intricate tasks, potentially understanding emotional cues and providing personalized service [29]. - The ongoing development of Qianlang's embodied intelligence architecture is expected to expand the operational boundaries of humanoid robots, making them integral to service industries [29][30].

麻省理工学院计算机科学与人工智能实验室（CSAIL）近期开发了一个新系统，为机器人控制提供了全新 的视角。该系统无需使用手工设计的模型或复杂的传感器阵列，而是允许机器人仅通过视觉来学习其身体 如何响应控制命令。这种名为"神经雅可比场"（NJF）的方法赋予机器人一种身体自我意识。这种建模与 硬件设计的分离可以显著拓展机器人的设计空间。在软体机器人和仿生机器人中，设计师通常会嵌入传感 器或加固部分结构，以便建模。NJF打破了这一限制。该系统无需机载传感器或设计调整即可实现控制。 设计师可以更自由地探索非传统的、不受约束的形态，而不必担心以后是否能够对其进行建模或控制。 3、 ADAM半人形机器人在肯尼迪航天中心航天飞机机舱内提供饮料 美国哥伦比亚大学的研究人员开发出一种名为Truss Link的机器人，它可以探测并与附近的机器人零件合 并，以填补缺失的部件。这种新型人造肌肉可以像人类肌肉一样运动，但强度是人类肌肉的17倍，机器人 的桁架连接件由磁力棒制成，可以扩展或从平面形状转变为3D结构以适应环境。它还可以添加其他机器人 的新部件或丢弃不再起作用的旧部件以提高其性能。在该团队发布的视频中，机器人与附近的一块物体 ...

机器人大讲堂获悉 ， 「加速进化」近日官 宣成功完成 A+轮融资。 本轮融资由北京市机器人产业发展投资 基金领投，北京市人工智能产业投资基金及博华资本联合跟投，义柏资本长期担任独家财务顾问。 其中， 北 京市机器人产业发展投资基金由京国瑞和首程控股共同管理，北京市人工智能产业投资基金由顺禧基金和启明 创投共同管理。 据了解，加速进化成立于 2023年， 核心团队源自清华火神队，在足式机器人算法与硬件领域拥有20余年技 术积累。 该公司专注于人形机器人、操作系统及开发工具等软硬件平台的研发，致力于推动具身智能技术的 商业化落地。今年6月初，加速进化刚刚完成A轮融资，而此次新一轮融资距离上一轮仅不到两个月，投资方 包括政府产业基金、财务投资机构及产业资本。加速进化为何能在短期内连续获得融资？ 在 2025年巴西RoboCup机器人世界杯中，加速进化的T1和K1人形机器人被多支人形组国内外强队选用， 包括清华火神、农大山海、清华TH-MOS、美国UT Austin Villa及德国Boosted HTWK等。 在成人组赛事 中，使用 T1机器人的清华火神队在成人组以5:2夺冠， 实现了 中国战队在该赛事 28年历史 ...

由于 2023 年度财务会计报告被出具无法表示意见的审计报告， ST 工智自 2024 年 5 月 6 日起便被实施 退市风险警示。 6 月 12 日晚， ST 工智（ 000584 ）发布公告称，公司于当日收到深交所送达的《关于江苏哈工智能机器 人股份有限公司股票终止上市的决定》，深交所决定终止公司股票上市。 公告显示， ST 工智股票自 6 月 20 日起进入退市整理期，退市整理期届满的次一交易日，深交所对公司股 票予以摘牌。 如今，哈工智能已经确定成为首个主板退市的机器人公司。 ▍ 哈工智能的成长与落寞 其实 ST 工智，在上市的 30 年中，经历了多次更名，其背后则是实控人和主营业务的更迭 ， 属于实打实 "改头换面" 。 1995 年 ， 一家名为蜀都 A 的公司成功上市，该公司以成都市蜀都大厦的物业为基础，兼具房地产开发、 商业贸易等业务。 2003 年， 江苏双良科技有限公司（以下简称"双良科技"）收购了蜀都 A 的 29.99% 股权。而后，经过资 产置换和资产收购，氨纶产业逐步成为公司的经营主业。 2004 年，其证券简称变更为舒卡股份。 2009 年 ，为了更好地发挥出"友利"的品牌优势， ...

近年来，人工智能在语言理解、视觉识别等非物理交互领域取得了显著进展，但在涉及实际物理操作的灵巧控 制方面，机器人的能力仍然相对薄弱。即使是简单的抓取、操控等任务，现有机器人系统的表现仍远不及人 类，甚至难以达到婴幼儿的灵活程度。 在灵巧手技术领域，目前尚未形成统一的技术路线标准。 2指、3指、4指和5指构型等各类设计方案，以及绳 驱、直驱等驱动方式，仍处于并行发展阶段。从产业化进程来看，灵巧手技术仍处于关键技术攻关和方案验证 阶段。 ▍ 以人手为天然范本，锚定仿人绳驱式五指 在机器人灵巧操作领域，仿人灵巧手的设计需要满足多方面的技术要求。 从功能需求来看，理想的灵巧手应 当具备高自由度运动能力、精确的力位控制、环境自适应性和快速响应等特性。 这些需求与人手的操作特性 高度吻合，也决定了五指仿生结构的技术合理性。 人类双手是长期进化的优化成果，其结构为灵巧手仿生设计提供了天然范本。 五根手指中，拇指可对握，与 四指协同完成抓、握、捏等动作；指节比例接近黄金分割（如中指三节长度比约为 1:0.6:0.4），在灵活性与 力量间达到平衡；每指三关节的设计实现毫米级操作精度；皮肤、肌腱与骨骼的配合，既保证发力能力，又具 ...