The internet was convinced China's newest humanoid was human. But did it fool our reporter when they got to see it up close. This is Iron, created by Xping, the Chinese Tesla competitor known for electric vehicles, flying cars, and robots that are so real they need to be sliced open to prove they aren't human.And when Iron was unveiled, the internet kind of lost its mind. Some people were convinced this was just someone in a suit. So naturally, X-ping CEO grabbed a knife and cut it open.Mashable reporter Am ...

2025深圳（国际）市容环境产业博览会与人工智能环卫机器人大赛近日同步收官。通过"展赛联动"创新 模式，活动实现4600万美元国际合作签约及1600台环卫机器人意向订单，总金额超3亿元。实景演示场 让市民亲身体验智慧城市治理，大赛则推动技术快速转化，构建起从产业展示到实景验证的完整闭环。 吸引海内外150余家优秀企业参展 本次博览会成为产业资源聚合的"强磁场"，实现了技术展示与商业落地的高效转化。博览会静态展示 区、动态体验区合计吸引海内外150余家优秀企业参展，展览总面积达50,000平方米，累计吸引超3万人 次参观，覆盖园林绿化、市容街区、环境卫生全产业链，深城交AI服务机器人、玉禾田坎德拉S300清 扫机器人等一批前沿新品集中发布，为行业发展把脉趋势。 国际经贸交流环节成效凸显，活动连接阿联酋、俄罗斯、哈萨克斯坦等"一带一路"共建国家资源，为中 国智慧城市治理方案出海搭建了重要桥梁，现场签署合作意向书8份，累计金额高达4600万美元，约 3.22亿元人民币。 国际经贸交流。 此外飞盘挑战赛、户外露营等配套活动人气高涨，"在公园就能体验未来生活"成为市民高频评价。 莲花山市民活动。 大赛通过多维度赛事设置 ...

转自：草原云 今年8月，国务院发布了《关于深入实施"人工智能+"行动的意见》，对人工智能与各行各业深度融合作 出战略部署，在《意见》的指引下，四川制造业正经历一场智能化变革。接下来，我们就跟随记者深入 生产一线，走进能源重装、消费电子等领域车间，感受AI赋能下四川制造业的新变化。 01 智能制造 精度与效率实现"双增" 在一座生产汽轮机叶片的工厂，记者看到，智能物流设备正快速移动，将叶片毛坯精准送达机器人臂 的"手中"，在这近万平方米的厂房区域中，维持运转的人员仅有10多人。 东方汽轮机叶片分厂厂长 任宗华：通过智能制造，让我们的人均效率达到了650%，质量水平也全部达 到99%以上，通过AI实时监测与管控，我们的精度已经达到0.03毫米级。 0.03毫米，相当于一张A4纸厚度的三分之一，精细度堪比一根头发丝。智能制造不仅提升了精度，也提 升了效率。 让机器人工作提质增效 智能车间里的高效率，离不开技术和工具的不断创新，从引进技术到创造技术，四川正通过自主研发的 AI技术，助力智能制造不断更新迭代。 长虹智慧显示（中国）MES工作室主任 冯楚贤：我们这条生产线通过MES系统和AI视觉检测深度融 合，以前我们的 ...

ADAM, the robot barista serving up shots of java at CESMartine ParisCES, the global innovation showcase that has long set the tone for the year ahead in tech, is returning to Las Vegas in January to explore the shift in AI from gimmicky features to the foundation of nearly every product category. LG and Samsung will be sharing their vision for the home of the future. Lenovo will be taking over The Sphere with an immersive keynote and experience. Sony Honda Mobility will be giving its next generation electri ...

在具身智能迈向真实世界应用的关键阶段，大规模、高质量、多平台兼容的机器人操作数据已成为制约技术突 破的核心瓶颈：一方面，双臂操作作为最贴近人类行为的"刚需"形态，正成为行业主流趋势，但受限于高昂的 采集成本与复杂的标注难度，相关数据极度稀缺；另一方面，现有数据集普遍存在真实场景覆盖不足、任务单 一、过度实验室化等问题，且大多仅适配特定或有限种类的机器人本体与构型，缺乏跨平台、跨本体的通用 性。 为破解这一难题，北京智源人工智能研究院牵头，联合 蚂蚁天玑实验室、银河通用、乐聚、软通天擎、松 灵、星海图、智平方、睿尔曼 等产业先锋，以及 清华大学、北京大学、斯坦福、伯克利、剑桥 等海内外顶尖 学 术 力 量 ， 共 同 打 造 并 发 布 了 RoboCOIN （ Bimanual Robotic Data COllection for INtegrated Manipulation） —— 全球"本体数最多、标注最精细、使用最便捷"的高质量双臂机器人真机数据集 。 破局三大挑战：格式混乱、质量参差、使用门槛高 当前具身智能数据正面临 "标准缺失、质控薄弱、工具分散" 三大核心痛点，严重制约行业发展： 为系统性解 ...

Core Concept - The article discusses a silent revolution in robotics, focusing on the integration of "mechanical metamaterials" to enhance robots' capabilities by allowing them to perceive, react, and adapt autonomously, merging central command with instinctive responses [1][4]. Group 1: Current Challenges in Robotics - Traditional robots operate on a "brain + body" model, where a central processor controls all functions, leading to limitations in flexibility and responsiveness [5][9]. - Current robotic systems are often heavy and rigid, making them unsuitable for delicate tasks that require safe interaction [6][9]. - The reliance on a central brain for every action creates bottlenecks, especially in unpredictable environments where rapid response is crucial [9][10]. Group 2: Mechanical Metamaterials - Mechanical metamaterials are defined as intelligent materials with programmable structures that can directly interpret and execute complex mechanical commands [7][11]. - The design principles of metamaterials include mechanics-inspired architectures, reconfigurable structures, and material-driven functionalities, which collectively enhance robots' physical capabilities [8][11][17]. Group 3: Innovations in Robotic Design - The first principle involves creating lightweight yet strong lattice structures that allow for adaptable flexibility and energy storage for rapid movements [11][12]. - The second principle utilizes origami and kirigami techniques to enable robots to dynamically change their shapes, enhancing their adaptability to various environments [13][15]. - The third principle integrates responsive materials that allow robots to sense and react to their surroundings, effectively merging sensory and motor functions [17][19]. Group 4: AI and Metamaterials - The integration of AI with metamaterials is seen as a key to overcoming current design challenges, allowing for efficient modeling and intelligent control of robotic systems [23][24]. - AI can assist in reverse engineering designs based on desired functionalities, improving the efficiency of creating complex metamaterial structures [24][26]. Group 5: Future Vision - The ultimate vision for metamaterial robots is a deep integration of embodied intelligence, where robots can operate more like biological entities, with distributed sensory and decision-making capabilities [27][28]. - Future robots are expected to function effectively in dynamic environments, with the ability to self-assemble and reconfigure into different forms as needed [28].

Core Viewpoint - The humanoid robot industry, which gained significant attention in 2023, is now facing a wave of bankruptcies, highlighting a stark contrast between expectations and reality in the market [11]. Group 1: Industry Overview - Multiple companies, including Datar and OneStar, have recently faced severe operational issues, including layoffs and cash flow problems, leading to their downfall [1][4]. - The humanoid robot sector has seen a surge in financing events, with 114 occurrences in the first five months of 2025, surpassing the total for 2024 [4][3]. - Despite the recent financing boom, the industry is experiencing a significant number of closures, reflecting the challenges of transitioning from concept to commercial viability [11][5]. Group 2: Reasons for Failures - The market for companion and social robots is limited, with functionalities easily replaced by smart speakers and apps, making it a high-risk area for failure [6]. - High-cost general-purpose humanoid robots face challenges due to complex technology, low margins, and a lack of clear market demand, leading to inevitable failures [7]. - Companies overly reliant on parent companies for support are at high risk, as strategic shifts or funding withdrawals can jeopardize their survival [8]. - Products with high hardware costs but low added value struggle to maintain cash flow, making it difficult for companies to sustain operations [9]. - Research prototypes lacking practical applications are unlikely to achieve profitability, leading to potential business failures [10]. Group 3: Notable Bankruptcy Cases - Anki, a consumer robotics company, declared bankruptcy in 2019 due to insufficient market demand for its high-cost interactive robots [17]. - Alphabet's Everyday Robots project was shut down in 2023 due to high costs and limited product applicability [19]. - K-Scale Labs, which focused on low-cost humanoid robots, ceased operations in November 2025 due to financing failures and cash flow issues [21]. - Rethink Robotics, known for collaborative robots, filed for bankruptcy in 2025 after failing to meet sales expectations and facing investor withdrawal [14]. - OneStar, established in May 2025, disbanded within five months due to internal conflicts with its parent company, Geely [37]. Group 4: Market Characteristics - Consumer-level companion and social robot companies often face high R&D costs and limited market size, making them vulnerable to failure [42][44]. - General service robots derived from research institutions struggle with unclear commercialization paths and high costs [46][48]. - Companies supported by parent firms may face existential threats if those firms withdraw support or shift focus [50][52]. - Startups often lack the financial stability to survive without continuous funding, making them susceptible to market fluctuations [54][56]. Group 5: Common Challenges - High R&D and production costs coupled with low margins are significant barriers to success in the humanoid robot industry [55][57]. - Insufficient market demand and unclear commercialization paths hinder the ability of humanoid robots to achieve profitability [58][59]. - Heavy reliance on external financing creates vulnerabilities, as seen in multiple cases where funding failures led to business closures [59][61]. - Strategic adjustments by parent companies can lead to rapid business failures for dependent startups [62][63].

Core Insights - The article discusses the investment strategies and insights of Brian Zhan, a partner at Striker Venture Partners, focusing on identifying potential unicorn companies in the AI era [1][2][3] Investment Strategies - Brian emphasizes the importance of recognizing specific patterns: top technical talent combined with problems deemed "almost impossible to solve" [1] - The investment in Reflection AI, which raised $2 billion and saw a 15-fold increase in valuation within six months, highlights the focus on reinforcement learning (RL) as a key breakthrough for making AI agents reliable [2] - The investments in foundational AI infrastructure companies like Lepton, Voyage, and LanceDB were based on the foresight that these would be essential for future AI applications [2] Robotics and AI - Brian views Skild AI as the "OpenAI of robotics," noting that the perception of robotics as a "capital graveyard" has changed due to advancements in foundational models and the availability of vast amounts of robotic data [3] - The emergence of general-purpose robots is seen as a paradigm shift, allowing for the development of useful robotic systems at significantly lower costs [3] AI for Science - The timeline for true AI for science is projected between 2025 and 2030, with current models beginning to perform genuine reasoning within scientific concepts rather than merely matching patterns in scientific texts [4] - The concept of "digital biological systems" is introduced, emphasizing the ability of models to understand biological pathways across multiple scales and navigate high-dimensional combinatorial spaces efficiently [5] Next-Generation Agents - Current AI agents lack several capabilities, including strong intelligence, robust multi-modal abilities, and effective self-improvement mechanisms [6] - The need for agents to build shared written records and effectively transfer knowledge is highlighted as a critical gap in their development [6] Investment Philosophy - Striker Venture Partners adopts a concentrated investment strategy, committing up to $30 million in only 10 companies, which fosters deep collaboration and alignment with founders [8] - The focus is on identifying opportunities that are not yet obvious, with a belief that the next billion-dollar company will emerge from a team currently operating under the radar [11] Advice for Founders - Founders are encouraged to act decisively when they identify a technological breakthrough that others have not yet recognized, as the best time to build transformative companies is often when the market has not yet defined the problem [12]

点击下方 卡片 ，关注" 具身智能 之心 "公众号 作者丨 Ting Huang等 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。 在四足机器人的视觉 - 语言 - 动作（VLA）领域，核心难题在于 弥合高层语义推理与底层动作控制的鸿沟 ——现有方法要么直接映射语言到动作导致稳定性差，要 么依赖隐式中间嵌入缺乏可解释性，难以在真实场景中稳健执行复杂任务。 北京大学等团队推出的 MobileVLA-R1 ，通过 "结构化思维链（CoT）推理 + 两阶段训练范式"，首次实现四足机器人在自然语言指令下的显式推理与连续控制，在 导航和操作任务中性能领先，为移动机器人的通用化落地提供了全新方案。 为什么需要重构移动机器人的 VLA 框架？ 当前四足机器人的 VLA 系统面临两大核心痛点，形成 "推理 - 控制脱节" 的困境： | 核心痛点 | 具体表现 | | --- | --- | | 语义 - 控制鸿沟 | 直接从语言映射到动作，缺乏中间推理环节，导致 ...

Ondas Holdings Inc. (NASDAQ:ONDS) is among the 10 Best Drone Stocks to Buy Right Now. On November 26, Lake Street analyst Max Michaelis lifted his price target on the stock to $10 from $9, while maintaining a Buy rating. Lake Street Lifts Ondas Holdings Inc. (ONDS)'s Price Target, Maintains Buy Rating Following Roboteam Acquisition Lake Street’s update follows the company’s announcement a day prior to acquire Roboteam in an $80 million agreement. The analyst anticipates Ondas Holdings Inc. (NASDAQ:ONDS) ...