Core Viewpoint - The Beijing Humanoid Robot Innovation Center has launched the first fully autonomous humanoid robot tour guide solution in China, which integrates various advanced technologies to facilitate diverse applications in exhibition halls, shopping malls, business presentations, and cultural tourism sites [1] Group 1: Technology and Innovation - The solution is based on the "Wisdom Opens Things" general embodied intelligence platform, which integrates autonomous guiding, human-like interaction, multi-machine scheduling, and global IoT linkage capabilities [1] - The core of the fully autonomous guiding capability is a closed-loop technology system built on the "perception-decision-execution" framework, driven by AI large models and a data-driven approach [1] - The collaboration of the embodied "brain," data-driven "small brain," and multi-modal perception system allows the robot to operate independently without human control throughout the guiding process [1] Group 2: Application and Future Prospects - The solution is expected to be widely applicable in various scenarios, including exhibition guiding, shopping assistance, business explanations, and cultural tourism [1] - Future applications will leverage multi-machine collaboration and comprehensive connectivity to enhance user experiences across different environments [1]

齐鲁晚报·齐鲁壹点 尚青龙 高雅洁 除了精密"大脑"，人形机器人推广的另一重要制约因素是协调流畅的行动能力。青岛钢铁侠科技有限公 司十年磨一剑，为机器人锻造最强"小脑"。以ARTROBOT系列大型双足仿人机器人为例，从初代仅能 双足行走，到如今第五代可完成端茶倒水、拧螺丝等复杂精细动作，每一步都凝聚着研发团队的心血。 "第一代ARTROBOT只有两条腿，我们的目标很简单，就是先实现稳定的双足行走。虽然面临诸多技术 难题，但团队坚持攻克。"青岛钢铁侠科技有限公司副总裁陈天祥介绍道，第五代ARTROBOT搭载的自 研"运动脑"，集成精密硬件与智能软件系统，在提升算力的同时降低功耗，为机器人的精准动作输出提 供核心支撑。"'运动脑'如同人类的小脑，体积不大却功能强大。它让机器人的每一个动作都更加精 准、流畅，极大便利了多自由度机器人的开发。" 康养领域的领跑布局 乐聚智家和钢铁侠作为引入的"生力军"，为青岛具身智能产业注入了活力。而青岛本土企业在智能机器 人领域，尤其在康养医疗领域已抢占先机。 山东卓业医疗研发的全球首台AI经皮穿刺手术导航机器人，集成3D结构光与动态感知技术，可实现亚 毫米级穿刺精度。"这台机器人 ...

Core Insights - The article discusses a teaser ad promoting "Unauthorized Silent Partners" of Nvidia, highlighting companies involved in breakthrough technologies that are not officially recognized by Nvidia [1][2] - The focus is on three companies that are positioned to benefit from Nvidia's advancements in quantum computing and robotics, suggesting significant investment opportunities [27][29] Quantum Computing Partners - The first two companies identified are involved in quantum computing, with Nvidia investing heavily in this area, indicating a shift from theoretical to practical applications [9][10] - The first silent partner is described as a "precision builder" of quantum intelligence infrastructure, with partnerships with major firms like Microsoft and Amazon, and a significant patent portfolio [13][14] - The second partner is a "hybrid pioneer" in quantum technology, backed by notable investors and government contracts, and is already offering commercially available products [22][23] Robotics and Autonomous Vehicles - The third company is linked to robotics, specifically autonomous trucks, which Nvidia sees as a major opportunity, with significant investments from top venture capital firms [29][32] - This partner has established collaborations with major companies like FedEx and Volvo, and is working on proprietary software for autonomous vehicles [34] - The article suggests that this company could be Aurora Innovation, which is developing a hub-and-spoke network for autonomous trucking and has made progress in commercial operations [35][36]

Core Insights - Hangzhou Yundeshuchu Technology Co., Ltd. has completed a C-round financing of over 500 million RMB, led by Zhaoshang International and Huaxia Fund, with participation from various strategic investors including China Telecom and China Unicom [1][4] - The financing aims to enhance product research and development, capacity expansion, and promote the commercialization of embodied intelligent robots [1][3] Company Development - The company has accelerated its development pace this year, launching the M20 quadruped robot and the DR02 humanoid robot, and establishing a pilot base for embodied intelligence [2][6] - The M20 robot is designed for complex terrains and hazardous environments, while the DR02 is an all-weather humanoid robot capable of real-time perception and intelligent decision-making [2][6] Strategic Initiatives - The establishment of the pilot base marks a significant step towards the industrial application of intelligent robots, creating a comprehensive ecosystem for testing, standardization, production, and application [2][6] - The company plans to deepen its dual-driven strategy of "independent innovation + industrial collaboration," focusing on core technology and production base development [6][7] Market Position and Growth - The company has expanded its business coverage to 34 provincial-level administrative regions in China and 44 countries and regions overseas, with a projected revenue growth of over 100% in 2024 compared to 2023 [4][7] - The company aims to launch a new consumer product by 2026, targeting a price point below 10,000 RMB, differentiating itself from existing products on major e-commerce platforms [5][6] Investor Confidence - The increased investment from both new and existing shareholders reflects growing confidence in the company's business model and commercial trajectory [4][7] - Investors have noted the company's impressive technological accumulation and product innovation capabilities, as well as its rapid commercialization progress [4][7]

截至10:24，机器人ETF（562500）下跌1.03%，报0.964元。盘中价格在探底后呈现企稳迹象，相对 便宜的"黄金坑"或已显现。持仓股方面，成份股表现分化，跌多涨少。73只持仓股中，11只上涨，62只 下跌。弘讯科技领涨3.15%，华东数控上涨2.44%，华昌达、东杰智能逆势飘红。巨轮智能下跌3.99%， 景业智能、固高科技跌幅居前，均跌超3%。 尽管盘面调整，但交投维持高热度。截至当前，成交额已达2.93亿元，换手率1.15%，显示出市场 多空博弈激烈，承接盘较为有力。值得注意的是，资金对机器人赛道长期价值高度认可，数据显示，机 器人ETF（562500）昨日单日"吸金"超1亿元，近期的调整或成为资金逆势加码、优化持仓成本的良 机。 每日经济新闻 （责任编辑：张晓波 ） 【免责声明】本文仅代表作者本人观点，与和讯网无关。和讯网站对文中陈述、观点判断保持中立，不对所包含内容 的准确性、可靠性或完整性提供任何明示或暗示的保证。请读者仅作参考，并请自行承担全部责任。邮箱： news_center@staff.hexun.com 华西证券表示，国内外科技巨头争相入局，人形机器人产业化提速，随着AI突破&政 ...

RoBoa是苏黎世联邦理工学院的衍生公司，由Alexander Kübler、Pascal Auf der Maur、Betim Djambazi与Nicolas Aymon共同创立。Alexander Kübler为公司首 席执行官，拥有苏黎世联邦理工学院机械工程硕士学位，曾担任该学校机器人实验室科学研究员。Pascal Auf der Maur为公司首席技术官，拥有苏黎世联邦 理工学院机器人硕士学位。Betim Djambazi为公司首席客户官，拥有苏黎世联邦理工学院机械工程硕士学位。Nicolas Aymon为公司首席运营官，拥有苏黎世 联邦理工学院机械工程硕士学位。 图源startupticker 据统计，2024年全球应急救援机器人市场规模约为6.82亿美元，2020-2024年CAGR约为12.55%。在各行业检测及救援行动中，狭窄密闭空间作业始终是一项 重大挑战。目前市场的主要解决方式为使用硬体机器人与无人机，但这些装备难以在蜿蜒曲折、尖锐或湿滑的环境中行动，在杂乱管道等危险或难以进入的 区域的作业效果大大受限，而人工检测救援等传统方法会导致高昂的中断成本和人员安全隐患。 图源RoBoa RoB ...

Core Insights - The article discusses the development of the Self-Referential Policy Optimization (SRPO) framework, which addresses the limitations of existing Visual Language Action (VLA) models in robotic tasks by enabling robots to learn from their own experiences without relying on external expert data [3][10][56]. Motivation and Contribution - SRPO aims to overcome the challenges of sparse reward signals in reinforcement learning, particularly in the VLA domain, by utilizing self-generated successful trajectories to provide progressive rewards for failed attempts [6][10]. - The framework eliminates the need for costly expert demonstrations and task-specific reward engineering, thus enhancing the efficiency of the learning process [10][12]. Technical Approach - SRPO collects trajectories generated during policy inference and categorizes them into successful and failed attempts, using a potential world representation to model behavior similarity [16][17]. - The framework employs a progressive reward mechanism based on the distance of failed trajectories to successful trajectory representations, allowing for a more nuanced evaluation of task progress [22][24]. Experimental Results - SRPO achieved a success rate of 99.2% in the LIBERO benchmark with only 200 steps of reinforcement learning, significantly outperforming traditional methods that rely on sparse rewards [29][30]. - In the LIBERO-Plus generalization tests, SRPO demonstrated a performance improvement of 167%, showcasing its robust generalization capabilities without the need for additional training data [31][32]. Efficiency and Real-World Application - The efficiency of SRPO is highlighted by its ability to improve success rates from 17.3% to 98.6% in long-term tasks with minimal training steps, outperforming other models in terms of training efficiency [36][39]. - The framework has been tested in real-world scenarios, showing significant improvements in success rates compared to supervised fine-tuning baselines [41][39]. Conclusion - SRPO represents a significant advancement in robotic learning, allowing for autonomous exploration and creativity by enabling robots to learn from their own successes and failures, thus paving the way for a new approach in VLA reinforcement learning [56].

编辑丨 量子位 点击下方 卡片 ，关注" 具身智能之心 "公众号 >> 点击进入→ 具身 智能之心 技术交流群 更多干货，欢迎加入国内首个具身智能全栈学习社区： 具身智能之心知识星球（戳我） ，这里包含所有你想要的! 让机器人听懂指令，精准导航再升级！ 深圳大学李坚强教授团队最近联合北京理工莫斯科大学等机构，提出视觉-语言导航 （ VLN ） 新框架—— UNeMo 。 通过 多模态世界模型 与 分层预测反馈机制 ，能够让导航智能体不仅可以看到当前环境，还能预测接下来可能看到的内容，并据此做出更聪 明的决策。 相比主流方法，UNeMo可大幅度降低资源消耗，在未见过的环境中导航成功率可达72.5%，尤其是在 长轨迹导航 中表现突出。 目前，该论文已入选AAAI2026。 以下是更多详细内容。 语言推理与视觉导航的"脱节困境" 作为Embodied AI的核心任务之一，视觉-语言导航要求智能体仅凭 视觉图像 和 自然语言 指令，在未知环境中自主完成目标导航。 而随着大语言模型 （ LLM ） 的兴起，基于LLM的导航方法虽取得进展，但仍面临两大关键瓶颈： 双模块协同打造"预判+决策"闭环 推理模态单一：现有方法仅 ...

点击下方 卡片 ，关注" 具身智能 之心 "公众号 编辑丨具身智能之心 本文只做学术分享，如有侵权，联系删文 >> 点击进入→ 具身智能之心 技术交流群 更多干货，欢迎加入国内首个具身智能全栈学习社区 ： 具身智能之心知识星球 (戳我) ， 这里包含所有你想要的。 让机器人「热面包」 在混乱桌面中快速找齐文具，还能精细处理不同形状、材质和尺寸的物品⚡️ 「把可乐放进蓝盘」 甚至推理出先用左臂，但遇障时换右手拿更快 从走路、跳舞到后空翻，动作模仿教会了机器人「怎么动」，而到端盘子、分拣水果、热食物等复杂操作时，机器人不能只模仿，更要识别复杂环境，理解「为什 么做」的任务意图，再转化为「动手这么做」的连贯操作。 人类的行动，一般都依托于上下文和意图，核心就在于推理。对机器人而言，尽管大规模互联网数据让GPT、DeepSeek等AI具备了不错的推理能力，但让AI在真实 物理世界里通过推理"准确动起来"，特别是处理多步骤长时序任务、模糊指令、未见过情景时，依然挑战重重。 尽管没见过这块面包，机器人通过推理识别它，推理出加热＝用微波炉，以及开门、拿起、放入、关门、旋钮、等待、取出……无需编程，全程推理完成！ 「整理文具 ...

格隆汇12月11日｜英思特在特定对象调研时表示，在机器人领域，公司已开展前瞻性的技术储备与产品 研发，并组建了专业的技术团队。目前，相关业务进展顺利，部分产品已实现小批量交付，同时也在与 部分机器人电机厂商进行样品测试与开发，为未来该领域的市场拓展奠定基础。 ...