根据最终协议，收购对价将根据Mentee交割前期权归属情况调整，交易已获Mobileye董事会及最大股东 英特尔的批准，因Mentee董事长Amnon Shashua同时担任Mobileye首席执行官，其已回避相关决策投 票。交割后，Mentee将作为独立事业部运营，保持核心团队完整性，同时受益于Mobileye的AI训练基础 设施。 美国当地时间1月6日，辅助驾驶与高等级驾驶自动化技术巨头Mobileye在拉斯维加斯消费电子展 （CES）上宣布，已就收购AI（人工智能）人形机器人公司Mentee Robotics（以下简称"Mentee"）达成 最终协议。据悉，本次收购对价总额约为9亿美元（现汇率约合 62.96 亿元人民币），其中包括约6.12亿 美元现金及至多约2620万股Mobileye A类普通股。收购预计于2026年第一季度完成。 Mobileye方面对《中国经营报》记者表示，Mentee在人形机器人领域取得了突破性的成果，Mobileye则 在驾驶自动化领域拥有深厚的技术积淀，并具备将先进的AI技术产品化的能力。此次收购有望推动物 理AI（Physical AI）在机器人与驾驶自动化两大分支的长 ...

本文来自微信公众号： 秦朔朋友圈 ，作者：悟00000空，题图来自：视觉中国 君子报仇，十年不晚 2026年国际消费类电子产品展览会 （简称CES） 最炸裂的新闻估计就是黄仁勋发布了英伟达开发的 自动驾驶平台Alpamayo，并免费开放，允许所有潜在用户利用该平台训练自家模型、研究开发相关 技术。 黄仁勋表示： "物理AI的ChatGPT时刻 （The ChatGPT moment for physical AI） 已到来 ——机 器开始理解、推理并在真实世界中行动。无人出租车是首批受益者之一。Alpamayo为自动驾驶汽车 带来推理能力，使其能够思考罕见场景，在复杂环境中安全驾驶，并解释其驾驶决策——这是安全、 可扩展自动驾驶的基础。" 这意味着特斯拉在自动驾驶方面的优势被其他车企追平甚至赶超的概率大大提高，因为英伟达的这个 平台有两大特点。 第一，它是首个具有推理能力、能够端到端融合处理多模态 （视觉、语言、动作，VLA） 数据的自 动驾驶平台。 也就是说，在这个平台上训练出来的自动驾驶模型有思考能力，能够处理罕见的意外情况，比如交通 信号灯故障。去年12月，谷歌旗下Waymo公司的自动驾驶出租车在旧金山 ...

Core Insights - The journey of David Moss across the United States using Tesla's Full Self-Driving (FSD) system demonstrates the potential of achieving fully autonomous driving without the need for LiDAR technology [2][17]. Group 1: Journey Overview - David Moss completed a 2-day, 20-hour journey covering 2,732.4 miles (approximately 4,397 kilometers) from Los Angeles to Myrtle Beach, South Carolina, without manual intervention [2][3]. - The trip faced various challenges, including low visibility fog, sudden rain, and complex urban traffic, yet the FSD system did not encounter any dangerous situations [5][6]. - Moss maintained an average speed of approximately 120 km/h, with a maximum speed of 136 km/h, and took about 12 hours of rest during the journey [5][6]. Group 2: FSD System Performance - The FSD system effectively managed tasks such as lane changes, traffic signal recognition, and parking at charging stations autonomously [5][6]. - Moss emphasized the need for constant attention to the road, as the system is still classified as a Level 2 driving assistance system, requiring driver supervision [6][9]. Group 3: Technology Comparison - Moss, originally a LiDAR salesperson, now supports Tesla's "pure vision" approach over multi-sensor fusion systems like Waymo's, arguing that achieving full autonomy does not necessarily require LiDAR [17][18]. - Tesla's FSD relies on a pure vision system that simulates human vision using cameras, while Waymo employs a multi-sensor approach that includes LiDAR and radar [18]. - The debate continues regarding the effectiveness of each system, particularly in extreme weather conditions and complex scenarios [17][18]. Group 4: Challenges to Commercialization - The successful completion of the journey does not equate to the readiness for commercial deployment of fully autonomous driving, as there are significant challenges to overcome [19][24]. - Key challenges include addressing long-tail risks, aligning technology with regulatory definitions, and the absence of a comprehensive regulatory framework for autonomous driving [21][23][24]. - Moss aims to highlight the value of autonomous driving technology, emphasizing its potential to benefit individuals who are unable to drive [24].

Group 1 - NVIDIA's CEO Jensen Huang predicts that a significant portion of cars will be autonomous or highly autonomous in the next decade, estimating that the number of autonomous vehicles could reach one billion, with every car being AI-driven [2][18] - At CES, NVIDIA introduced the Alpamayo series of autonomous vehicle models, which utilize a visual-language-action (VLA) model based on chain reasoning to enhance the development of safe, reasoning-based autonomous vehicles [2][18] - The first vehicle equipped with NVIDIA's technology is expected to be on the road in the U.S. in the first quarter of the year [2][18] Group 2 - In China, the Ministry of Industry and Information Technology announced the first batch of L3 conditional autonomous vehicle licenses, with Changan Deep Blue SL03 and Arcfox Alpha S6 being the first to receive approval, marking a significant step towards commercial application [3][19] - The L3 level of autonomous driving allows the system to perform all driving operations, with human intervention required only when requested by the system [20][22] Group 3 - The global autonomous taxi market is projected to grow from $4.43 billion in 2025 to $188.91 billion by 2034, while China's potential market is expected to increase from $39 million in 2025 to $67.59 billion by 2035 [12][28] - Major automotive manufacturers and ride-hailing platforms are likely to benefit from the growth of the autonomous taxi industry, including BYD, Geely, Great Wall Motors, SAIC, Xiaomi, and Didi [12][28] Group 4 - Autonomous driving technology is becoming an essential application of artificial intelligence, significantly impacting traffic safety, efficiency, and transportation options [9][25] - The performance and stability of autonomous driving systems are expected to improve with advancements in 5G, cloud computing, and the Internet of Things, leading to more breakthroughs and market applications [13][29]

Domestic News - The Ministry of Industry and Information Technology and seven other departments have issued an implementation opinion to accelerate the integration of artificial intelligence in industrial mother machines and industrial robots, focusing on developing new-generation AI numerical control systems and conducting tests for intelligent connected vehicles [7] - Hefei's 14th Five-Year Plan emphasizes the growth of emerging industries, particularly in sectors like "chip, display, automotive, and integration," aiming to enhance the scale and advantages of leading industries, including smart connected new energy vehicles [8] - Geely Auto has obtained an L3-level autonomous driving road test license in Hangzhou, which is the largest in terms of area and mileage in the country, allowing access to over 7,000 traffic data points [9] - Dongfeng Motor plans to start localized production of passenger vehicles in Turkey within the year, targeting a market with an annual sales volume of approximately 1.4 million vehicles [10] - Li Auto has launched a total of 3,900 Li Supercharging stations across 286 cities in China as of January 4, 2026 [11] - BAIC and Horizon Robotics have established a joint venture, Zhiyu Technology, focusing on core technologies for intelligent driving [12] - Baidu's Apollo Go has received the first full unmanned testing license in Dubai, marking a significant milestone in autonomous driving [13] - Intramco has partnered with Leap Motor to apply its charging products in Leap Motor's future European factory for electric vehicles [14] International News - Turkey's automotive market is projected to reach a record high of 1.37 million units in 2025, reflecting a year-on-year growth of 10.5% despite high taxes and tightening financing conditions [16] - Ford plans to launch a new driver assistance system by 2028, which will allow drivers to completely disengage from monitoring while driving [17] - Mercedes-Benz will introduce both fuel and electric versions of the next-generation S-Class, with separate platforms for each, indicating a strategic shift in their electric vehicle lineup [18] - Mobileye intends to acquire the humanoid robot manufacturer Mentee Robotics for $900 million, aiming to enhance its capabilities in autonomous driving and robotics [19] Commercial Vehicles - Henan Province will exempt hydrogen fuel trucks from tolls on expressways until the end of 2027, and electric trucks will receive a 50% toll discount [21] - Foton's Oumark has launched the "Zhi Nuan Core" thermal management system for electric trucks, addressing winter operational challenges with advanced temperature control technologies [22] - The Long-distance company has introduced its 3.0 generation product matrix, focusing on AI integration and high-efficiency solutions for various commercial vehicle segments [23] - Chongqing has released policies to support the development of intelligent connected new energy commercial vehicles from 2025 to 2027, including incentives for R&D and financing [24]

马斯克在回复一位网友讨论特斯拉FSD与英伟达Alpamayo的差距的帖子时称："要实现安全、无人监督 的自动驾驶，大约需要100亿英里的训练数据。现实中的长尾问题超级复杂。" 该网友在帖子中写道："对于许多人来说，区分演示和产品是很困难的，这是可以理解的。在技术方 面，一些产品的演示效果很好，已经接近最终的生产就绪状态。对其他人来说，演示和产品之间的差距 是巨大的。非工程师往往低估了跨越这一鸿沟所需的过程。 有些问题比他们的演示所显示的要困难得多，'全自动驾驶'就是其中之一。使用FSD，从引人注目的演 示到成品的过渡是非常困难的。问题空间是由一组有效的无界边缘情况定义的，而进展则是由打磨长尾 决定的。 这就是为什么，鉴于我的背景，我很容易忽视NVIDIA的Alpamayo平台。有些人认为我的观点是傲慢 的。我不同意。 你必须了解特斯拉为了接近最终目标所经历的整个过程。这需要数年持续不断的努力。即使从FSD v12 的架构重置开始，也需要大量的工作来系统地、逐次迭代地减少长尾问题，如今已发展到v14。 马斯克在回复一位网友讨论特斯拉FSD与英伟达Alpamayo的差距的帖子时称："要实现安全、无人监督 的自动驾 ...

Core Insights - Nvidia has launched an open-source AI model platform, Alpamayo, aimed at enhancing autonomous driving capabilities, which is expected to accelerate the penetration of high-level autonomous driving and Robo-X services [2][3] Group 1: Nvidia's Innovations - Nvidia's CEO Jensen Huang introduced the open-source VLA (Vision-Language-Action) model platform, Alpamayo, at CES 2026, along with simulation tools and an open dataset containing over 1,700 hours of driving data, creating a comprehensive autonomous driving ecosystem [2] - Alpamayo, as a reasoning VLA model, can progressively solve complex problems and generate reasoning traces similar to human thought processes, enabling safe driving in complex environments and explaining safety decisions [2][3] Group 2: Market Trends and Projections - The current stage of Robotaxi services is approaching a break-even point per vehicle, with leading companies like Loongbo and Pony.ai achieving or nearing this milestone, indicating a potential increase in revenue and profitability for Robotaxi businesses [3] - The deployment costs for unmanned logistics vehicles have dropped below 100,000 yuan, making them a cost-effective alternative to manual labor, with predictions of nearly 40,000 units shipped in China by 2025 and 100,000 units by 2026 [3] Group 3: Opportunities for Suppliers - Nvidia's development of a global L4-level autonomous driving and Robotaxi ecosystem includes partnerships with various operators and suppliers, suggesting a growing demand for high-level autonomous driving components such as smart driving chips, LiDAR, domain controllers, and electronic control systems [3] - The expansion of Nvidia's autonomous driving ecosystem is expected to create ongoing opportunities for suppliers of both hardware and software in the autonomous driving sector [3] Group 4: Investment Recommendations - The launch of the open-source autonomous driving model Alpamayo is anticipated to benefit automotive companies, component suppliers, and mobility service providers, with specific investment targets identified in the report [4]

Core Insights - Nvidia has launched the open-source AI model platform Alpamayo for autonomous driving at CES 2026, which is expected to enhance the penetration of advanced autonomous driving and Robo-X services [1][2] - The platform includes simulation tools and an open dataset with over 1,700 hours of driving data, aiming to create a comprehensive autonomous driving ecosystem [1] - Nvidia's CEO Jensen Huang announced that the NVIDIA DRIVE AV system will first be integrated into the Mercedes CLA, with plans for a self-driving taxi service testing in 2027 [1] Group 1: Technology and Innovation - Alpamayo, as a reasoning VLA model, can solve complex problems and generate reasoning traces similar to human thought processes, enabling safe driving in complex environments [2] - The introduction of Alpamayo is expected to accelerate the iteration of autonomous driving technology and promote the deployment of advanced autonomous driving and Robotaxi/Robovan services [2] Group 2: Market Trends and Projections - The Robotaxi sector is approaching a break-even point for individual vehicles, with leading companies like Luobo Kuaipao and Pony.ai achieving or nearing this milestone [3] - The deployment costs for unmanned logistics vehicles have dropped below 100,000 yuan, making them a cost-effective alternative to human labor [3] - Predictions indicate that the shipment volume of unmanned delivery vehicles in China will reach nearly 40,000 units in 2025 and is expected to double to 100,000 units in 2026 [3] Group 3: Ecosystem Development - Nvidia is building a global L4-level autonomous driving and Robotaxi ecosystem, collaborating with various operators and suppliers in the industry [4] - The demand for advanced autonomous driving components such as intelligent driving chips, LiDAR, domain controllers, and electronic control systems is anticipated to grow rapidly as the ecosystem expands [4] Group 4: Investment Opportunities - The launch of the Alpamayo platform is expected to benefit automotive companies, component suppliers, and mobility service providers [5] - Key investment targets include component suppliers like Jingwei Hirain (688326.SH), Desay SV (002920.SZ), and vehicle manufacturers such as XPeng Motors (09868) and Pony.ai (02026) [5]

Group 1 - The core viewpoint of the article is that Yunda Technology (300440.SZ) is involved in the research and production of L4 level autonomous driving logistics robots and solutions for trunk logistics with autonomous trucks [1] Group 2 - The company has a stake in Yangsi Technology, which focuses on the development of advanced autonomous driving technologies [1]

Core Viewpoint - The concept of "data closed loop" in the autonomous driving industry is still largely limited to small internal loops within algorithm teams, rather than achieving the grand vision of a comprehensive system that directly solves problems through data [1]. Group 1: Definition of "True Data Closed Loop" - A "true closed loop" must meet three levels: automated problem discovery, quantifiable and reviewable solution effects, and a comprehensive trigger system that integrates real-time and historical data [4][5]. - The ideal state involves a system that can automatically classify issues, route them to the appropriate teams, and assist in developing trigger rules, thereby reducing reliance on manual processes [5]. Group 2: Current Industry Practices - Many companies' so-called "data closed loops" are more accurately described as "data-driven development processes with some automation tools," primarily limited to the perspective of individual algorithm teams [8]. - Typical workflows are often module-level and algorithm-focused, lacking a system-wide perspective [9]. Group 3: Reasons for Lack of True Closed Loops - The starting point for many companies is a "passive closed loop," where problems are identified reactively rather than through automated data analysis [10]. - Attribution of issues is often difficult, as multiple interrelated factors contribute to the same phenomenon [12]. - The data-to-solution chain often stops at data-to-model, failing to address real-world problems effectively [16]. Group 4: Data Closed Loop Practices - The company has developed a more aggressive approach to data closed loops, treating data as a product and metrics as primary citizens [24]. - The overall strategy involves quantifying real-world pain points and using triggers to convert these into actionable data [25]. Group 5: Trigger Mechanism - The trigger mechanism is designed to be lightweight and high-recall, ensuring that significant events are captured without overwhelming the system [32]. - Once a trigger is activated, it generates a micro log that is uploaded for further analysis, leading to more detailed data collection if necessary [35]. Group 6: Unified Trigger Framework - A unified trigger framework using Python allows for consistent implementation across vehicle data mining, cloud data analysis, and simulation validation [50]. - This framework enables non-technical team members to participate in writing rules, thus democratizing the process of data analysis [54]. Group 7: Distinction Between World Labels and Algorithm Labels - The company maintains two types of labels: world-level labels that describe objective physical conditions and model-level labels that depend on algorithm performance [61]. - This distinction is crucial for effective data analysis and problem-solving in the autonomous driving context [61]. Group 8: Use of Generative and Simulation Data - Generative data is primarily used to address long-tail scenarios that are difficult to encounter in real life, but real data remains essential for evaluation and validation [67]. - The company emphasizes the importance of filtering data through structured labels before applying vector retrieval methods to ensure efficiency and accuracy [64].