机器人赛道迎来利好政策。 8月11日，杭州市司法局发布公开征求《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》）意见的公告，通过制度设计规范发展秩序、优化产业生态、强化要素保障，全面推动具身智能机器人产 业实现高质量发展。 值得一提的是，《条例》提到，促进具身智能机器人在工业制造、农业生产、医疗健康、教育培训、特种作 业、公共安全等重点领域场景的应用推广。 重磅发布 8月11日消息，杭州市司法局发布公开征求《杭州市促进具身智能机器人产业发展条例（草案）》）意见的公 告，将具身智能机器人产业纳入全市政策支持框架，通过制度设计增强政策合力，进一步规范产业高质量发展 路径。 《条例》指出，面对当前研发训练需求和未来具身智能机器人的广泛应用需求，要求加强网络和算力基础设施 规划布局，建立多元智算供给服务体系，提升算力资源利用效率，降低算力使用成本。 《条例》明确核心技术主攻方向，聚焦具身智能"大脑"（具身智能大模型）、"小脑"（运动控制系统）及"本 体"（机器人核心部件和整机）三大核心以及专用芯片等"卡脖子"环节，具体来看： 在"大脑"研发方面，支持开展具身智能模型的研发和应用，提升多模态感知融合 ...

在2025世界机器人大会（下称"WRC2025"）上，越疆科技发布具身智能人形机器人DOBOTAtom第二代产品。越疆早期凭借创新的四轴机械臂迅速占领市场， 此后随着市场需求的变化，越疆将战略重心转向更高附加值的六轴协作机器人领域，2023年起正式启动人形机器人研发。 越疆做人形机器人是基于哪些考量？人形机器人的生态未来会如何发展？带着这些问题，证券时报记者在WRC2025期间专访了越疆首席科学家郎需林。 从工业机器人到人形机器人 证券时报：越疆大概什么时候开始研发人形机器人？为什么考虑做人形机器人？越疆人形机器人的特色和亮点在哪儿？ 郎需林： 我们开始的不算早，但我们选对了路径。越疆从2023年起正式启动人形机器人研发，但我们更早就在做底层准备——从控制器、驱动器到关节模组的自 研能力，为人形机器人铺设了底层"脊椎"。 之所以进场，是因为我们看到AI模型涌现带来的控制、认知、交互能力突破，让"人形"不再是外形模仿，而是"人类行为的具身再现"。 我们的Atmo系列强调"可量产、可部署"，具备闭环控制系统、轻量化关节模组，以及适配VLA等大模型的多模态感知与任务泛化能力，这让它不仅能走路，还能工 作。 证券时报 ...

Core Insights - The article discusses the launch of the G0 model by Xinghai Map, which enables robots to autonomously and continuously perform tasks like bed-making with minimal fine-tuning [1][3] - The G0 model integrates a large-scale open-world dataset with a dual-system architecture to enhance the robot's adaptability in unstructured environments [3][4] Dataset Overview - The Galaxea Open-World Dataset is the first high-quality dataset collected in real human environments, covering 50 types of settings and totaling 500 hours of high-quality mobile operation data [7][9] - The dataset includes over 150 tasks, 1600+ objects, and 58 skills, emphasizing diversity and complexity in task distribution [10][12] - Data collection utilized a unified robotic platform, ensuring consistent action space and perception inputs across various tasks and scenarios [8][9] Model Architecture - The G0 model employs a dual-system structure, separating high-level reasoning and low-level action execution into G0-VLM and G0-VLA modules [18][19] - G0-VLM processes natural language task instructions and breaks them down into executable subtasks, while G0-VLA executes actions at a high frequency [19] Training Strategy - The training of the G0 model is divided into three phases: cross-platform pre-training, single-platform pre-training, and task-specific post-training [21][23] - This phased approach enhances the model's performance and adaptability to complex skills, particularly in long-sequence tasks [21][23] Performance Evaluation - The G0 model outperformed the π0 benchmark model in various tasks, demonstrating significant advantages in object manipulation and language instruction following [22][30] - The model achieved over 50% improvement in accuracy compared to baseline models, particularly after specialized training [32] Conclusion - The combination of the Galaxea Open-World Dataset and the G0 dual-system VLA model provides a scalable and high-fidelity approach for training and deploying embodied intelligence [33] - The open-sourcing of data and models aims to bridge technological gaps and accelerate the transition of embodied intelligence from laboratory innovations to societal applications [33]

1. Report Industry Investment Rating - Not mentioned in the provided content 2. Core Viewpoints of the Report - Last week (2025.08.04 - 2025.08.08), the market rebounded with the support of the banking sector. The CSI 300 index rose 1.23% and the CSI 500 index rose 1.78% weekly. However, market structural instability increased, featuring faster theme rotation, intensified differentiation, and more obvious poor - performance characteristics of hot stocks [2][11]. - The robot industry is shifting from "showcasing technology" to "practical use", driven by technological progress and cost reduction. The military industry will remain a key focus in the market due to factors like performance repair, approaching parade, and geopolitical conflicts [3][12]. - In the short - term, the performance of AI chain's computing power and vertical application companies is highly related to the release rhythm of domestic and foreign models and their horizontal comparison. In the long - term, it is affected by actual commercialization progress and implementation [4][13]. - Many active equity funds have announced restrictions on purchases to ensure stable operation and protect the interests of fund holders in the context of significantly improved excess returns of active funds [4]. 3. Summary by Relevant Catalogs 3.1 Weekly Market Observation 3.1.1 Equity Market Review and Observation - Market trends: The market rebounded last week with the support of the banking sector. Some indices reached new rebound highs, but structural instability increased. The market hotspots shifted to military and robot sectors, while innovative drugs and the North American AI industry chain declined. Since August, incremental funds have changed, with the accelerated inflow of margin trading funds pushing small - cap stocks to new highs. The current market trading volume is moderately increasing but still far from the previous high. The margin trading balance exceeded 2 trillion yuan last week, bringing potential risks [11]. - Industry trends: The robot industry is moving towards practical use, with technological progress and cost reduction driving development. The military industry will be a market focus due to factors such as performance repair, approaching parade, and geopolitical conflicts. The performance of the AI chain is affected by model releases in the short - term and commercialization in the long - term [3][12][13]. 3.1.2 Public Fund Market Dynamics - Many active equity funds have announced purchase restrictions. For example, China Europe Fund announced restrictions on China Europe Medical Innovation and China Europe Science and Technology Innovation Theme funds to ensure stable operation and protect the interests of fund holders [4][14]. 3.2 Active Equity Fund Index Performance Tracking 3.2.1 Active Stock - Picking Fund Index - Index positioning: 15 actively managed equity funds are selected each period, equally weighted. Core positions select funds based on performance competitiveness and style stability in value, balanced, and growth styles, and the style distribution is balanced according to the CSI Active Stock - Type Fund Index [16]. - Performance: It rose 1.55% last week and has accumulated an excess return of 12.73% since its establishment [6]. 3.2.2 Value Stock - Picking Fund Index - Index positioning: Select 10 funds of deep - value, quality - value, and balanced - value styles based on multi - period style classification. The value style includes different types of value investment [18]. - Performance: It rose 1.69% last week and has accumulated an excess return of - 3.29% since its establishment [6]. 3.2.3 Balanced Stock - Picking Fund Index - Index positioning: Select 10 relatively balanced and value - growth style funds based on multi - period style classification. Fund managers balance stock valuation and growth and consider cost - effectiveness in both stock selection and industry allocation [20]. - Performance: It rose 1.70% last week and has accumulated an excess return of 7.85% since its establishment [6]. 3.2.4 Growth Stock - Picking Fund Index - Index positioning: Select 10 funds of active - growth, quality - growth, and balanced - growth styles based on multi - period style classification, aiming to capture the double - click opportunities of performance and valuation in high - growth companies [23]. - Performance: It rose 1.05% last week and has accumulated an excess return of 20.49% since its establishment [6]. 3.2.5 Pharmaceutical Stock - Picking Fund Index - Index positioning: Select 15 funds based on the intersection market value ratio of fund equity holdings and the representative pharmaceutical index, and evaluate them through multiple indicators such as relative benchmark index winning rate, drawdown level, style stability, and overall performance competitiveness [23]. - Performance: It fell 1.90% last week and has accumulated an excess return of 22.32% since its establishment [6]. 3.2.6 Consumption Stock - Picking Fund Index - Index positioning: Select 10 funds based on the intersection market value ratio of fund equity holdings and representative consumption - related indices, and evaluate them through multiple indicators [26]. - Performance: It rose 3.34% last week and has accumulated an excess return of 16.68% since its establishment [6]. 3.2.7 Technology Stock - Picking Fund Index - Index positioning: Select 10 funds based on the intersection market value ratio of fund equity holdings and representative technology - related indices, and evaluate them through multiple indicators [30]. - Performance: It rose 1.48% last week and has accumulated an excess return of 19.53% since its establishment [6]. 3.2.8 High - end Manufacturing Stock - Picking Fund Index - Index positioning: Select 10 funds based on the intersection market value ratio of fund equity holdings and representative high - end manufacturing indices, and evaluate them through multiple indicators [30]. - Performance: It rose 1.85% last week and has accumulated an excess return of - 1.33% since its establishment [6]. 3.2.9 Cyclical Stock - Picking Fund Index - Index positioning: Select 5 funds based on the intersection market value ratio of fund equity holdings and representative cyclical indices, and evaluate them through multiple indicators [34]. - Performance: It rose 2.88% last week and has accumulated an excess return of - 0.57% since its establishment [6].

去年至今，人形机器人成为投资领域出现的 "高频词"。 作为未来产业的前沿领域，人形机器人正成为全球科技和产业竞争的新赛道。据高盛预测，到 2035 年全球市场规模有望突破千亿美元。特 斯拉 CEO 埃隆·马斯克展望，长期来看人形机器人需求量可能达 100 亿台。 随着人形机器人产业加速落地，其对能源系统的挑战正日益凸显。 锂电池作为核心动力源，面临更高能量密度、更快充电速度、更高安全性 和更优温域适应性的多重考验， 这也促使电池企业从 "汽车"快速走向"人形"的新战场。 据亿纬锂能消息， 近日，维他动力与亿纬锂能正式达成机器人业务深度合作。 双方将通力协作，基于用户需求和机器人动力电池技术发展趋势，推进具身智能产品化量产的进一步发展。 01 牵手维他动力 亿纬锂能开发人形机器人专用电池 资料显示，今年 5 月， Vbot 维他动力已完成天使轮融资，由知名投资机构达晨创投领投，初心资本、联想创投、渶策资本跟投，高鹄资本 担任独家财务顾问。 成立半年，维他动力融资规模达 3 亿元，完成首款智能伴随机器人的投产及规模化的前置准备，将在 9 月启动制造工厂， 10 月开启量产级 早鸟内测，今年底完成产品上市，并开启海 ...

Core Viewpoint - Nvidia is diversifying its investments in the field of embodied intelligence by collaborating with multiple Chinese robotics companies, similar to its previous investment in OpenAI [1][6]. Group 1: Nvidia's Technological Advancements - Nvidia has delivered its first batch of Jetson Thor chips to Galaxy General, showcasing impressive performance improvements, including a 7.5 to nearly 10 times increase in computing power compared to the previous generation Jetson Orin [2][30]. - The Jetson Thor chip also features a 3.5 times improvement in performance per watt and a 10 times increase in I/O throughput, catering to high-bandwidth perception needs [30][31]. Group 2: Collaborations and Industry Trends - Companies like Yushu Technology and Galaxy General are closely collaborating with Nvidia, focusing on different approaches to robotics, with Yushu emphasizing "motion-first" and Galaxy General adopting an "intelligence-first" strategy [4][6]. - Other companies such as Alibaba Cloud and various robotics firms are also engaging in deep collaborations with Nvidia, indicating a broad industry trend towards integrating advanced AI and robotics technologies [6]. Group 3: Market Predictions and Future Developments - The market for humanoid robots is predicted to grow exponentially, with Galaxy General's CTO forecasting a tenfold increase in market value every three years [6][15]. - The future of robotics is envisioned to allow for easy assembly of robots similar to computer assembly, driven by advancements in AI and hardware simplification [8][10]. Group 4: Product Innovations and Applications - Galaxy General has launched several innovative robots, including a humanoid robot capable of performing complex tasks and a robot dog designed for industrial applications, showcasing significant advancements in design and functionality [11][14]. - The company has also introduced a 24-hour unmanned pharmacy solution and plans to deploy unmanned retail stores, demonstrating the commercial viability of its robotic solutions [23][24]. Group 5: Nvidia's Role in Robotics - Nvidia is focusing on three types of computers essential for robotics: the robot body computer, AI factory computer, and simulation computer, each playing a critical role in the development and deployment of robotic systems [28]. - The Isaac platform, which includes the Jetson Thor, is designed to support the development of intelligent agents in the physical world, enhancing the capabilities of robots [29][32].

作者 I 杨文静 报道 I 投资界PEdaily 一群清华校友齐聚深圳南山。 从一场融资路演说起。 8月7日上午，深圳大学城国际会议中心，"X-Da y "西丽湖路演社清华校友项目专场正式上演。投资界从现场看到，本次活动汇聚了21 家清华校友创业企业，遍布人工智能、具身智能、高端装备、医疗服务等领域；同时台下超百家投资机构身影活跃。 当天上午，路演拉开帷幕—— 首先上场的是 大寰机器人 。2016年成立，大寰总部位于深圳南山，三位联合创始人师从学术泰斗戴建生教授，其中此次上台路演的创 始人兼CEO孙杰是清华大学机械自动化专业校友。 专注于灵巧手机器人，大寰已完成多轮融资，身后集结了IDG资本、字节跳动、君联资本、启明创投等明星投资方身影。 现场一位00后女性创业者令人印象深刻，她是 Ze r ith零次方 联合创始人李宜哲。身处当下火爆的具身智能技术研赛道，零次方孵化于 清华机器人实验室，20 25年初正式成立，至今已推出双足人形机器人Z1和轮式人形机器人H1。截至目前，零次方已经完成3轮天使轮 融资，投资方包括包河创投、同创伟业、力合科创、水木基金、滨湖金投集团、国耀资本等我们熟悉的投资机构。 值得一提的是 ...

新浪科技了解到，这一实际运营场景依托银河通用在零售领域的深耕布局。WRC 2025前夕，银河通用 还发布了为城市生活量身打造的人形机器人智慧零售解决方案——银河太空舱：每台太空舱最小仅需9 ㎡空间，却可由机器人全天候自主运营，售卖饮品、小食、文创、药品等高频消费品类，兼容冷藏/冷 冻等多种商品形态。单舱预计日均服务人次可达2000、订单量达500单，周边人流预估提升30%-40%， 有效激活区域活力。 新浪科技讯 8月11日下午消息，2025年世界机器人大会（WRC 2025）期间，北京银河通用机器人有限 公司（简称"银河通用"）携具身大模型机器人Galbot参展，展示了覆盖商业售卖、工业及城市服务等领 域的端到端具身大模型落地应用。 在会客厅的机器人市集，银河通用展示了自主售卖"大世界"文创产品的机器人。观众扫码下单后， Galbot能够自主完成从商品定位、取货到交付的全流程操作，整个过程无需人工干预。忙碌的机器人穿 梭其间，精准服务每一位顾客，生动展现了具身智能在商业零售中的实际应用能力。 在工业领域的汽车零件SPS分拣工作中，Galbot攻克了传统机器人难以突破的复杂难题——面对多层堆 叠且错位的零件、 ...

Core Viewpoint - The 2025 World Robot Conference held in Beijing focuses on making robots smarter and more intelligent, showcasing advancements in industrial and humanoid robots, particularly by Zoomlion, which aims to provide a new model for the intelligent transformation of the manufacturing industry [1] Group 1: Company Innovations - Zoomlion is one of the earliest adopters of robotic technology in the engineering machinery sector, having introduced programmable industrial robotic arms in 2006 for welding components [1] - Since 2019, Zoomlion has extensively utilized adaptive robots with perception and mobility capabilities, including vision and tactile recognition robotic arms and autonomous mobile robots (AMRs) [2] - The company has integrated over 300 intelligent manufacturing production lines in its Changsha facility, employing more than 2,000 industrial robots for a fully traceable and scheduled production process [2] Group 2: Humanoid Robots Development - Zoomlion has developed three humanoid robots, including two wheeled and one bipedal version, and has deployed dozens of these robots in factory operations for tasks such as logistics, assembly, and quality inspection [3] - The humanoid robots are equipped with visual systems for component inspection and are part of a training facility with nearly 100 workstations for large-scale data collection [3] Group 3: Future Vision - The company envisions a future factory model where industrial and humanoid robots work collaboratively, evolving from large-scale agile manufacturing to personalized custom production [3] - Zoomlion plans to continue integrating industrial internet, big data, artificial intelligence, and robotics to expand the boundaries of intelligent manufacturing and reshape the development paradigm of the engineering machinery industry [3]

Investment Rating - The report maintains an "Outperform" rating for the industry, particularly focusing on humanoid robots and AI infrastructure [4][8]. Core Insights - The humanoid robot sector is rapidly evolving, with significant advancements showcased at the 2025 World Robot Conference, highlighting the industry's potential for growth and innovation [3][16]. - The report emphasizes the importance of supply chain positioning and value capture in humanoid robotics, recommending a focus on key suppliers and companies with strong market positions [4]. - AI infrastructure is expected to see sustained capital expenditure growth driven by increasing demand, particularly benefiting sectors like gas turbines and chillers [4]. Industry Dynamics - The 2025 World Robot Conference opened in Beijing, featuring over 200 companies and 1500 exhibits, showcasing advancements in humanoid robotics and their applications [16]. - The report notes the rapid development of the domestic robot industry, with improvements in capabilities such as sorting, handling, and performing tasks like music and shopping [3]. - The low-altitude economy is progressing with infrastructure initiatives, including the successful completion of the first offshore oil platform supply flight using drones [27]. Company Dynamics - Key companies highlighted for humanoid robotics include: - **Joint suppliers**: Hengli Hydraulic, Huichuan Technology, and Bluesky Technology for joint modules [4]. - **Dexterous hands**: Zhaowei Electromechanical and Leisai Intelligent [4]. - **Reduction gears**: Green Harmonic and Jindi Co., Ltd. [4]. - AI infrastructure companies to watch include Hanbell Precise Machinery, Ice Wheel Environment, and Yingliu Co., Ltd. [4]. Key Company Earnings Forecasts and Valuations - The report provides earnings forecasts and investment ratings for several companies, including: - Green Harmonic (688017.SH): Outperform, with a projected EPS of 0.50 for 2025 [8]. - Huichuan Technology (300124.SZ): Outperform, with a projected EPS of 2.01 for 2025 [8]. - Bluesky Technology (688360.SH): Outperform, with a projected EPS of 0.99 for 2025 [8]. - The report also highlights the performance of companies in the humanoid robot sector, noting significant market movements and growth potential [15].