引言 过去一年，具身智能成为机器人产业最热的关键词之一。人形机器人、服务机器人、特种机器人密集发布，但 一个现实问题也越来越清晰地浮出水面： ▍ 从 " 卖机器人 " 到 " 卖大脑 " ：产业逻辑正在发生变化 智能能力深度绑定硬件 算法围绕单一场景优化 产品更新依赖重新设计整套系统 机器人看起来更 " 聪明 " 了，但离真正的自主智能，仍然存在结构性差距。 近日， Robo Summit 机器人发展论坛（第二期）暨首届朝阳智能机器人生态大会在北京朝阳举办。千诀科 技创始人兼 CEO 高海川受邀发表主题演讲，并与银河通用创始人兼 CTO 王鹤、智平方副总裁邱巍等嘉宾共 同启动 " 朝阳区智能机器人生态伙伴建设计划 " ，围绕具身智能的技术演进与产业协同展开深入探讨。 在演讲中，千诀科技创始人兼 CEO 高海川分享了一个颇具产业争议的判断 —— 机器人智能的下一阶段竞争，不在"哪种形态跑得更像人"，而在"机器人大脑是否真正独立、通用、可演 化"。 这一判断，正在重新划分机器人产业的技术路线。 长期以来，机器人产业的主流商业模式是 整机驱动 ： 但随着应用场景快速扩张，这种模式的边界越来越明显： 每进入一个新行业 ...

Core Insights - The article discusses the integration of Large Language Models (LLMs) with Behavior Trees (BT) to enhance robotic task execution and adaptability in the presence of external disturbances [1][2][12]. Group 1: LLM and BT Integration - LLMs are utilized to interpret user commands into behavior trees that include task goal conditions [2]. - The combination of LLMs and BT allows for fewer calls to LLMs while managing external disturbances through an action database [2][12]. - A human-in-the-loop learning mechanism is proposed to refine the knowledge generated by LLMs, ensuring safety and adaptability in robotic operations [5][7]. Group 2: Human-in-the-Loop Learning Mechanism - The mechanism involves designing a context for LLMs that includes prompt engineering, manipulation primitives (MPs), and an action database [5]. - User interactions guide LLMs to correct and enhance the generated action knowledge, which is then added to the action database after user confirmation [7][12]. - The generated action knowledge consists of preconditions, postconditions, and a set of MPs, implemented in BT format [7]. Group 3: Task Evaluation and Performance - Eight tasks were designed to evaluate the proposed method, categorized into three difficulty levels: Easy, Medium, and Hard [9]. - The proposed method achieved a success rate of over 80% across the tasks, significantly outperforming baseline methods that lacked human interaction [12]. - The adaptability of the generated action knowledge was tested against external disturbances, achieving a success rate exceeding 70% [14]. Group 4: Generalization and Future Improvements - The generated action knowledge demonstrated good generalization capabilities, with success rates over 70% for certain tasks involving new objects [17]. - However, some tasks had success rates below 40% due to the inapplicability of MPs parameters to new objects, indicating a need for fine-tuning before application [17]. - Overall, the proposed human-in-the-loop learning mechanism enhances robotic learning performance, enabling robots to complete tasks and respond to external disturbances effectively [18].

Core Viewpoint - The article discusses the launch of the robot rental platform "Qingtian Rent," which aims to create a new era in the robot rental industry by establishing a value co-creation and sharing mechanism, facilitating the accessibility of humanoid robots to the public [2][4]. Summary by Sections Platform Overview - "Qingtian Rent" integrates users, rental companies, content developers, and equipment manufacturers to enhance the overall value of the robot rental industry, defining a new era of RaaS (Robot as a Service) [4]. - The platform's core innovation is transforming high-threshold robot usage scenarios into a convenient rental model, similar to shared power banks, addressing industry pain points such as high operational costs and complex collaboration [4]. Strategic Goals - The "1234 Strategic Plan" aims to achieve partnerships with over 10 manufacturers, 200 premium service rental companies, 3,000 content creators, and 400,000 rental customers by 2026 [4]. - The ultimate belief is to leverage intelligent machines to create unlimited productivity, making efficient and flexible robot services permeate various industries [4]. Market Potential - The robot rental market has surpassed 1 billion yuan this year, with expectations to reach at least 10 billion yuan next year as manufacturers ramp up production and market demand increases [9][10]. - The anticipated growth is driven by the transition from novelty use to regular, scalable applications, similar to the evolution of ride-hailing services [10]. Usage Scenarios - The platform supports 13 major usage scenarios, including corporate events, exhibitions, retail promotions, and community security, indicating a broad potential for robot applications [11]. Open Platform Approach - "Qingtian Rent" is an open and inclusive platform, welcoming all robot manufacturers to join, regardless of their production capabilities, to match market demand with service providers [12]. Pricing Strategy - The platform aims to establish a competitive pricing system that satisfies both clients and rental service providers, ensuring it remains the most cost-effective robot rental platform nationwide [14].

企查查APP显示，近日，浙江创极星机器人有限公司成立，注册资本5000万元，经营范围包含：服务消 费机器人制造；服务消费机器人销售；工业机器人制造；智能机器人的研发等。企查查股权穿透显示， 该公司由宁波方正全资持股。 ...

Core Viewpoint - The article introduces the ManualVLA model, a unified VLA model designed to enhance robotic manipulation and task execution by integrating planning and action generation into a single framework, addressing challenges in long-duration tasks that require precise final state definitions [2][5][10]. Group 1: Research Background and Challenges - Recent advancements in VLA models have significantly improved robotic scene understanding and generalization, yet challenges remain in coordinating high-level planning with precise operations for long-duration tasks like LEGO assembly and object rearrangement [7]. - Two main challenges are identified: the need for precise operations to align with predefined final configurations and the integration of long-term planning with fine-grained control while maintaining generalization capabilities in diverse real-world environments [7][9]. Group 2: ManualVLA Method Description - ManualVLA allows the model to generate its own instruction manual and execute actions based on it, breaking down complex long-duration tasks into controllable and interpretable short phases [12][19]. - The model employs a Mixture-of-Transformers (MoT) architecture, integrating a planning expert that generates multimodal operation manuals and an action expert that executes the actions based on these manuals [5][15]. - The ManualCoT reasoning mechanism combines explicit and implicit paths to influence action generation, ensuring a high degree of coordination between manual generation and action execution [16][20]. Group 3: Experimental Results - In real-world tasks, ManualVLA demonstrated a significant improvement in success rates, achieving an average success rate increase of approximately 32% compared to the latest baseline methods [28]. - The model's performance in generating intermediate target images was validated with metrics such as PSNR (e.g., 2D LEGO assembly at 29.01) and MAE (e.g., 2D LEGO assembly at 3.23), indicating high fidelity and accuracy in predicting target object positions [23][27]. - ManualVLA outperformed state-of-the-art methods in simulation tasks, achieving a 70% average success rate, surpassing the previous best of 63% [31]. Group 4: Ablation and Generalization Experiments - Ablation studies confirmed that all modalities of information in the instruction manual (text, images, UV coordinates) and the implicit CoT reasoning are essential for solving long-duration, goal-specific operational tasks [33]. - ManualVLA exhibited robust generalization capabilities under varying backgrounds, object shapes, and lighting conditions, maintaining high task success rates even in unseen scenarios [36].

Tesla Trade - Ark Invest sold 60,715 shares of Tesla across multiple ETFs, with a total transaction value of approximately $29.7 million, as the stock closed at $488.73 [2] - Concerns regarding Tesla's financial prospects were raised by Gerber Kawasaki, particularly about achieving GAAP profitability, although the company's market share in the EV sector remains strong [3] Palantir Trade - Ark Invest reduced its position in Palantir by selling 47,309 shares, valued at approximately $9.2 million, with the stock closing at $193.98 [4] - BofA Securities expressed optimism about Palantir's growth, citing strong momentum in its U.S. commercial business and a recent two-year $448 million government order [5] Shopify Trade - Ark Invest sold 33,164 shares of Shopify, valued at approximately $5.6 million, with a closing price of $169.67 [6] - Shopify's recent Winter 2026 Edition showcased expanded capabilities in AI and merchant tools, reinforcing confidence in its long-term growth according to JPMorgan [7] WeRide Trade - Ark Invest purchased 520,697 shares of WeRide, valued at approximately $4.7 million, with the stock closing at $8.97 [8] - WeRide launched public Robotaxi rides in Dubai, marking progress towards a fully driverless rollout planned for early 2026 [9] Other Key Trades - Ark Invest acquired 78,793 shares of CRISPR Therapeutics across ARKG and ARKK [10] - Intellia Therapeutics saw an acquisition of 187,880 shares via ARKG and ARKK, while Rocket Lab sold 232,425 shares through ARKQ and ARKX [12]

以下文章来源于刺猬公社 ，作者刺猬公社编辑部 刺猬公社 . 互联网内容行业观察与研究 来源丨刺猬公社（ID： ciweigongshe ） 作者丨 园长 编辑丨 陈梅希 图源丨Midjourney 北京后厂村，距离"互联网十字路口"不远的某座写字楼里，一间教室大小的房间内，几十个工人分成 两批，一组在桌上用螺丝刀和扳手拼装轴承模组，一组在旁边的测试区，用示波器和万用表对已经组 装好的总成做检测。 这不是什么小作坊，而是 2025 年的科技创业风口——人形机器人产业的生产环节。 从咖啡到汽车，我参观过各行各业数不清的智能工厂，通常是产品科技含量越高，需要人手工操作的 环节越少，流水线越标准规范，工人和流水线几乎融为一体。以此类推，创造"具身智能"的地方应该 充满了各种黑科技。 但情况并不是我事先想象的那样， 这里没有流水线，也没有除了金属工件冷冻机之外的大型设备，组 装规范全靠人手一本翻到起毛边的 A4 纸手册 ...... 它更像一间大学里自动化专业的实验室，或者一个忙忙碌碌的家电维修部。为了看看当前的机器人产 业究竟发展到什么程度，我报名了某个具身智能企业的外包兼职，本意是想在流水线上，看清这个行 业的冰山 ...

Summary of Key Points from the Conference Call Industry Overview - The focus of the conference call is on the **Robotics industry** in **North America** and the convergence of various technologies within this sector [1] Core Insights and Arguments - **Convergence of Technologies**: The term "convergence" is emphasized as a key theme, indicating the merging of AI agents with both digital and physical realms, as well as the integration of data and manufacturing processes [2][4] - **AI and Robotics Development**: AI chatbot firms are expanding their capabilities by forming robot teams and hiring manufacturing experts, highlighting a dual-purpose approach where consumer devices are also valuable in defense applications [6][6] - **Public and Private Sector Alignment**: There is a noted convergence of strategic goals between the private sector and public sector, particularly in the technological and military domains of the United States and China [6] - **Innovations from Major Companies**: - **Tesla** is filing patents for roof-integrated satellite antennae, indicating a push towards innovative technology integration [6] - **META** is utilizing facial data to train robots, showcasing advancements in AI and machine learning [6] - **Expansion of Market Segments**: Technology firms are increasingly entering industrial Total Addressable Markets (TAMs), while industrial firms are expanding into technology TAMs, indicating a blurring of industry lines [6] - **Elon Musk's Perspective**: Musk has acknowledged that his companies are trending towards convergence, reinforcing the idea that various sectors are becoming interconnected [6] - **Advancements in Autonomous Vehicles**: The development of autonomous cars is seen as a catalyst for advancements in low altitude robots (LARs), suggesting a significant overlap between automotive and robotics technologies [6] - **Interconnected Systems**: The relationship between brain-computer interfaces (human-to-machine) and humanoid robots (machine-to-human) is highlighted, indicating a future where these technologies may intersect [6] - **Investor Interest**: There is a growing convergence of investor interest in both public and private markets, reflecting a broader trend in the investment landscape [6] Additional Important Insights - **Data Collection**: Consumer devices equipped with cameras may soon be used to collect data for training robots, indicating a shift in how data is utilized in AI development [7] - **Space as a Connector**: The concept of space is described as the "connective tissue" of the embodied AI ecosystem, suggesting that spatial technologies will play a crucial role in future developments [7] - **Military Applications**: Autonomous cars are categorized as military-grade AI, indicating their potential use in defense and security applications [12] This summary encapsulates the key points discussed in the conference call, focusing on the convergence of technologies within the robotics industry and the implications for various stakeholders.

Morgan Stanley does and seeks to do business with companies covered in Morgan Stanley Research. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of Morgan Stanley Research. Investors should consider Morgan Stanley Research as only a single factor in making their investment decision. For analyst certification and other important disclosures, refer to the Disclosure Section, located at the end of this report. Downloaded by Neil.Wang@trowepr ...

Group 1 - The A-share market saw a collective rise in the three major indices on December 23, with sectors such as lithography machines, Hainan, insurance, and precious metals leading the gains, while sectors like horse racing, retail, Tianjin Free Trade Zone, and education experienced declines [1] - The largest AI ETF in the Shanghai market, AI ETF (515070), rose by 0.47% during the session, with holdings like Beijing Junzheng surging by 5.69%, and other notable gainers including Deep Sanda A, Cambricon Technologies, Amlogic, and Lexin Technology [1] - The "Qingtian Rental" platform, the first open robot rental platform in China, was launched at the National Robot Rental Ecological Summit on December 22, covering 50 core cities and linking over 600 service providers, with rental prices ranging from 200 yuan to over 10,000 yuan [1] Group 2 - According to Guotai Junan Securities, the humanoid robot industry is entering a critical mass production phase, with hardware companies becoming the core driving force. The period from 2022 to 2025 is seen as the first phase, transitioning products from laboratories to industrialization, while the second phase starting in 2025 will focus on large-scale production [1] - The number of registered humanoid robot companies in China is expected to reach 104 by 2024, marking a year-on-year increase of 104%. From January to July 2025, there have been over 101 financing events totaling more than 26 billion yuan, indicating a high level of enthusiasm in the capital market [1] - The AI ETF (515070) tracks the CS AI Theme Index (930713), selecting component stocks that provide technology, basic resources, and application end stocks for artificial intelligence, focusing on the midstream and upstream of the AI industry chain, often referred to as the "brain" creators of robots [2]