现在， 来自 瑞士洛桑联邦理工学院（ EPFL） 、美国 杜克大学 等机构 的 研究人员 不仅破解了这个秘密， 还造出了一条能在真实河流中游泳的机器鱼。 昨日 ，这项研究登上了全球机器人领域顶刊 《 Science Robotics》 的封面。 这项研究 模拟了斑马鱼视觉运动行为的神经回路 ， 他们 创建了名为 simZFish的虚拟仿真系统，并最终在 80厘米长的物理机器人ZBot上实现了验证。 更厉害的是，这套系统不仅能在实验室环境下重现斑马鱼的行为，还能在瑞士 Chamberonne河的急流中自 主导航，展现出了从实验室到真实世界的泛化能力。 鱼是怎么在湍急的河流中保持位置不被冲走的？ 这个看似简单的问题，背后涉及极其复杂的神经计算过程。 ▍ 复刻斑马鱼视觉 -运动神经通路 要理解这项研究的突破性，得先了解斑马鱼的 视动反应（ Optomotor Response, OMR） 。 当水流将鱼向后推时，鱼眼中看到的景物会向前移动。鱼的大脑检测到这种视觉运动后，会控制身体向前游动 来补偿位移，从而保持在原地。这个看似简单的行为，实际上涉及复杂的神经计算。 研究团队使用 物理引擎 Webots 构建了 sim ...

为精准传递 2025 年第二届中关村具身智能机器人应用大赛 核心信息，解答参赛者在赛制细节、报名流程及 政策支持方面的疑问，帮助参赛团队高效备赛， 10 月 14 日， 2025 年第二届中关村具身智能机器人应用 大赛 专项宣讲会隆重召开 。本次大赛是继 2024 年中关村仿生机器人大赛成功举办后，承接世界人形机器 人运动会国际成果升级打造的特色赛事，自 7 月 15 日报名通道开启 以来，已吸引全球多地企业、高校及科 研团队踊跃参与。 本次宣讲会汇聚了政府部门、产业园区、金融机构、企业及高校的众多重要嘉宾，共同为大赛赋能。政府领 导、园区负责人、金融机构代表及赛事承办方围绕赛事背景、产业支持、赛制规则等关键内容展开详细解读， 为参赛者搭建起清晰的备赛指引框架。 ▍ 海淀区筑牢产业根基政策资金护航赛事 中关村科学城 管理委员会 产业促进四处副处长郑雪介绍，海淀区作为全国人工智能创新核心区域，已形成深 厚的 具身 智能产业生态：聚集 300 余家创新企业（含 24 家人形机器人整机企业）、 21 所开设相关专业 的院校，及一批顶级科学家，构建 " 大脑、小脑、本体 " 全产业链，并 入选工信部机器人产业中小企业 ...

Group 1: Company Listings and Financing - Yunji Technology officially listed on the Hong Kong Stock Exchange, issuing 6.9 million shares at a price of HKD 95.6 per share, raising approximately HKD 590 million (about RMB 541 million) [3] - Since its establishment in 2014, Yunji Technology has completed 8 rounds of financing, with investors including Alibaba, Tencent, and Lenovo, achieving a post-investment valuation of RMB 4.08 billion in its last round in December 2021 [3] - Suzhou Junyue Robot Co., Ltd. completed a B+ round of financing, with funds aimed at increasing investment in new product development and global market expansion [16] Group 2: Product Launches and Orders - ZhiYuan Robotics launched the new generation industrial interactive robot, ZhiYuan Spirit G2, which has already secured orders worth several hundred million yuan [6] - UBTECH Technology announced a procurement agreement with a well-known automotive technology company for humanoid robots worth RMB 32 million, adding to a previous order of RMB 250 million, bringing the total contract amount for the Walker series to nearly RMB 500 million [9] Group 3: Industry Developments - RoboChallenge, the world's first large-scale physical robot benchmark evaluation platform, was launched to test robots on various tasks, bridging the gap between simulation testing and real-world deployment [12] - The ZhiYuan Spirit G2 is designed with high-performance motion joints and advanced spatial perception systems, supporting rapid learning and deployment across multiple scenarios [6]

地球表面 71% 被海洋覆盖，这片孕育生命的蓝色疆域不仅蕴藏着海量自然资源，更是调节全球生态的关键枢 纽。然而，超过 90% 的海洋区域仍处于人类探索的 "盲区"，加之人类活动对海洋环境的影响日益加深，对 深海进行低干扰、长时程、高适应性的监测，已成为海洋科学与生态保护的迫切需求。 研究团队提出了一种全新的驱动 /变形系统，核心是一种称为"后屈曲缺口板"（PBNP）的创新结构。 这种 设计模仿了蝠鲼的胸鳍，通过巧妙的机械结构将微小的线性运动转化为大幅度的鳍片拍动。 传统水下探测设备却难以突破环境局限：依赖螺旋桨等刚性部件的机器人，虽推力强劲，却在复杂流场、 狭 窄空间 或脆弱生态区 "水土不服"，容易卡在礁石缝隙中。而仿生机器鱼凭借柔软可变形的身体和仿生泳姿展 现出优势，仍面临游动模式单一、深海极端温度及非结构化环境适应性差等难题。机器鱼既能够在狭窄空间稳 定穿行、又能够在开阔水域欢快地游动，是 仿生机器鱼设计的核心挑战 。 而浙江大学研发的这款仿生机器鱼，恰恰打破了这一困境。 它跳出传统水下探测装置的设计框架，通过创新 的驱动 /变形系统，为复杂海洋环境下的低干扰探测与生态监测，开辟了全新路径。 这项研究， ...

刚刚启幕的第 138届广交会机器人服务专区，元化智能展台展示了一场海拔4000米的手术影像——在西藏比如县，一台锟铻®国产机器人在机械臂精准引导下， 辅助医生完成膝关节置换手术。 当全球医疗机器人领域仍在争论"中国能否造出可靠产品"时，创立6年多的元化智能给出了答案：从东部沿海到西部高海拔区域，从三甲医院到县级卫生院，全国 上百家医院完成了5000例临床手术， 证明了中国医疗机器人的高可靠性。 ▍ 高原上的实战： 5000例手术定义"可靠"新标准 元化智能科技（深圳）有限公司成立于 2018 年，定位高端专科智能医疗装备的创新研发和精密智造。公司成立6年 多 来，以全骨科手术辅助机器人为龙头，医 疗手术机器人产品覆盖骨科、消化科等多科室临床应用，是国内唯一拥有骨科手术机器人全部核心部件自主研发能力的公司。 在西藏那曲比如县人民医院的手术室，一台锟铻 ®正在工作。术前通过CT影像完成三维建模，系统自动规划截骨 方案和手术 路径。术中， 锟铻实时引导 操作， 在 关键步骤提供力控约束与路径导航。 "以前在高原做膝关节置换，最怕解剖结构变异导致截骨偏差，现在 机器人 能动态修正，误差控制在亚毫米内。 机器人不 仅保 ...

Core Viewpoint - Cloudwalk Technology (2670.HK) has emerged as a significant player in the robotics service industry, achieving a remarkable opening price increase of 49.37% on its debut, driven by strong market interest and a unique business model focused on hotel services and beyond [1][4][5]. Group 1: Market Logic Behind Subscription - The introduction of the Hong Kong Stock Exchange's 18C rule in 2023 aimed to attract hard-tech companies, but the number of listings has been limited until Cloudwalk's successful debut, which reignited interest in this sector [3]. - Cloudwalk's public offering saw an unprecedented oversubscription of 5657.2 times, with 263,000 retail investors participating, indicating a strong market appetite for innovative tech companies [4][5]. Group 2: Cloudwalk's Business Model and Growth - Cloudwalk has established itself as the only publicly listed company globally that has achieved large-scale commercialization of service robots, particularly in the hotel sector, where it holds a dominant market share of 13.9% [7]. - The company has expanded its service to over 34,000 hotels, completing 500 million service instances annually, showcasing its operational scale and efficiency [11][12]. Group 3: Expansion into New Markets - Cloudwalk is diversifying its business beyond hotels into high-value sectors such as healthcare and manufacturing, with a reported 79.1% increase in new contracts in non-hotel scenarios [15][16]. - The company has also made strides in international markets, with overseas revenue increasing from 8% in 2022 to 15% in 2025, aiming for over 30% in the next three years [16]. Group 4: R&D Investment and Technological Edge - Cloudwalk invests significantly in R&D, with 23.4% of its revenue allocated to this area in 2024, resulting in 985 patents that cover essential technologies in navigation, control, and AI decision-making [17][19]. - The company's proprietary "Service Intelligent Agent Autonomous Closed-Loop Learning System" allows robots to learn from real-world applications, enhancing their operational efficiency [19]. Group 5: Financial Performance and Profitability Outlook - Cloudwalk's revenue has shown a compound annual growth rate of 23.4% from 2022 to 2024, with a notable increase in gross profit margin from 24.3% to 43.5% during the same period [20][22]. - The company aims to achieve breakeven by 2026, with a clear path to profitability supported by a stable hotel business and rapid growth in new sectors [22][23]. Group 6: Investment and Market Confidence - Cloudwalk has raised over 1.2 billion yuan through 11 funding rounds, attracting investments from major players like Tencent, Alibaba, and Ctrip, reflecting strong market confidence in its business model and growth potential [25][26]. - The company's focus on a profitable business model, unlike many peers that rely on heavy losses, positions it favorably in the competitive landscape of service robotics [28][30]. Group 7: Leadership and Vision - Founder Zhi Tao, a notable figure in a male-dominated industry, emphasizes the mission of using robots to alleviate human labor in repetitive tasks, maintaining a long-term focus on commercial applications rather than consumer products [30][31]. - Cloudwalk's listing is seen as a milestone for the Chinese service robotics industry, demonstrating the potential for technology companies to achieve global competitiveness and profitability [31][32].

该成果以 " PneuGelSight: Soft robotic vision-based proprioception and tactile sensing "发表在国际权威期刊《 IJRR 》上 ▍ PneuGelSight 的 结构设计与制造工艺 PneuGelSight 是一种集成摄像头与内部照明的气动软体手指，能够在执行抓取任务时 同步实现高精度的本 体感 知 与触觉感知 。其结构分为两部分：背面采用 3D 打印的波纹管形态，易于在气压驱动下伸展；内侧 则由更厚、更硬的铸造硅胶板构成，不易伸展，从而在充气时促使手指向内弯曲，便于协同抓取物体。 软体气动机器人以其优异的柔顺性和自适应能力，在高端制造、精密装配和人机协作等工业场景中展现出巨大 潜力。然而，其传感系统的发展仍是一个核心瓶颈。 与结构明确、关节数有限的刚性机器人不同，软体机器人表面几乎每一个点的形变都会影响整体形态与功能， 导致其状态空间维度大幅升高。再加上软材料本身具有的非线性与复杂力学行为，使得建模与控制更为困难。 此外，高度可变形的结构也限制了传统刚性电路和电子元件的集成，而这些正是常规机器人传感系统的核心组 成部分。 为应对这些 ...

在未来的月球上，可能会出现一种奇怪的景象 ——一颗硕大的金属球，咕噜咕噜地滚下陨石坑，看起来像个 失控的健身球，但它其实是NASA背后的"新宠"：一台专为极端地形打造的球形探测机器人。 它的名字叫 ——RoboBall。它 既 不是玩具， 也 不是艺术装置，而是一台能自我平衡、能 "滚着下山"、还 能带着传感器执行任务的高智商胖球。 前不久， 来自德州农工大学（ Texas A&M University）的RAD实验室 成功研发了 RoboBall， 并 让这颗 球，重新定义 "月球漫步"的方式。 ▍探月这事，车轮已经不够用了 自从 NASA公布"Artemis"计划，人类就对月球南极蠢蠢欲动——那里可能藏着水冰，也藏着一堆"麻烦"。 月壑深不见底、坡度动辄几十度、尘土软得像面粉。在这种地形里，传统月球车就像开 SUV去冲沙——谁都 知道下去不一定上得来。 于是， RAD实验室决定换个思路：既然车轮搞不定，那不如——让整个机器人变成一个轮子。 听起来像是脑洞，但这正是 RoboBall的魔法： 它是一个充气的球，里面藏着一个摆锤 ， 通过摆锤前后摆动，它能改变重心，重心一偏，球就开始滚 ， 再 加上内部传感器 ...

1、 小米登记第三代人形机器人CyberOne作品著作权 天眼查App显示，近日，北京小米机器人技术有限公司登记"第三代人形机器人CyberOne"作品著作权，作 品类别为美术。该公司成立于2023年4月，法定代表人为曾学忠，注册资本约5556万人民币，由北京小米 智能科技有限公司、北京屹唐创欣创业投资中心（有限合伙）共同持股。公开资料显示，CyberOne（中 文名：铁大）是小米集团于2022年8月11日发布的首款全尺寸人形仿生机器人，主要应用于家庭护理与陪 伴场景。今年2月，小米开始推进该机器人在自有制造产线的分阶段落地。从行业趋势看，科技巨头纷纷 布局人形机器人，预示着通用机器人产业化进程加速。 2、 杭州率先立法护航具身智能机器人产业 10月13日，《杭州市促进具身智能机器人产业发展条例（草案）》（以下简称"《条例（草案）》"）经杭 州市政府常务会议审议通过，下一步拟提交杭州市人大常委会审议。这将是全国范围内首部针对具身智能 机器人产业的地方性法规，标志着杭州在布局未来产业、构建人工智能发展新优势方面迈出关键一步，以 制度创新为产业高质量发展赋能。《条例（草案）》系统构建了促进产业发展的整体框架。在技 ...

Core Insights - The discussion focuses on bridging the Sim2Real gap in robotics, emphasizing the importance of simulation in training robots to operate effectively in the real world [2][4][10] Group 1: Key Participants and Context - Madison Huang, NVIDIA's head of Omniverse and physical AI marketing, made her first public appearance in a podcast discussing robotics and simulation [1][2] - The conversation featured Dr. Xie Chen, CEO of Lightwheel Intelligence, who has extensive experience in the Sim2Real field, having previously led NVIDIA's autonomous driving simulation efforts [2][9] Group 2: Challenges in Robotics - The main challenges in bridging the Sim2Real gap are identified as perception differences, physical interaction discrepancies, and scene complexity variations [4][6] - Jim Fan, NVIDIA's chief scientist, highlighted that generative AI technologies could enhance the realism of simulations, thereby reducing perception gaps [6][7] Group 3: Importance of Simulation - Madison Huang stated that robots must experience the world rather than just read data, as real-world data collection is costly and inefficient [7][9] - The need for synthetic data is emphasized, as it can provide a scalable solution to the data scarcity problem in robotics [9][10] Group 4: NVIDIA's Technological Framework - NVIDIA's approach involves a "three-computer" logic: an AI supercomputer for processing information, a simulation computer for training in virtual environments, and a physical AI computer for real-world task execution [10][11] - The simulation computer, powered by Omniverse and Isaac Sim, is crucial for developing robots' perception and interaction capabilities [11][12] Group 5: Collaboration with Lightwheel Intelligence - The partnership with Lightwheel Intelligence is highlighted as essential for NVIDIA's physical AI ecosystem, focusing on solving data bottlenecks in robotics [15][16] - Both companies share a vision for SimReady assets, which must possess real physical properties to enhance simulation accuracy [16][15] Group 6: Future Directions - The live discussion is seen as an informal introduction to NVIDIA's physical intelligence strategy, which aims to create a comprehensive ecosystem for robotics [18] - As collaboration deepens, it is expected to transform traditional robotics technology pathways [18]