今天看到知乎一篇比较有意思的文章，根据海外绿神的一些分析判断特斯拉并非One Model，主要观点如下： 原文链接：https://zhuanlan.zhihu.com/p/1996714717922738540 据自动驾驶之心的了解，国内一段式端到端大都是One Model为主要模型，承担主要功能（95%），会增加一些小模型辅助做优化。从严格意义上来说，算不上真正 的"端到端"，但One Model本身是具备完整功能的。 分享这篇文章主要是想提供一个新的角度了解特斯拉FSD，欢迎大家理性讨论。 作者 | 刘延@知乎 编辑 | 自动驾驶之心 原文链接： https://zhuanlan.zhihu.com/p/1996714717922738540 点击下方 卡片 ，关注" 自动驾驶之心 "公众号 戳我-> 领取 自动驾驶近30个 方向 学习 路线 >>自动驾驶前沿信息获取 → 自动驾驶之心知识星球 本文已获转载授权，转载请联系原文作者 1、特斯拉并非One Model，而是近200个小场景模型的组合。 根据海外绿神对于固件的反向分析可以看到， 1）HW4有110和189两套模型组合，其中 ——节点 A 包含 ...

Core Insights - The Chinese autonomous driving industry is experiencing significant breakthroughs in large-scale operations, with companies like Pony.ai and WeRide achieving fleet sizes exceeding 1000 vehicles, marking a transition from "technical validation" to "scaled service operations" [1][7] Company Developments - Pony.ai announced a fleet size of 1159 vehicles as of December 2025, while WeRide reached 1023 vehicles by January 2026, both entering the "thousand-vehicle era" [1][3] - Pony.ai's CEO emphasized the importance of vehicle density in urban areas for profitability, stating that reaching a fleet size of 1000 vehicles allows for a shift from loss-making to at least break-even operations per vehicle [3] - WeRide reported that its fleet in Abu Dhabi is approaching single-vehicle break-even, while Pony.ai achieved single-vehicle profitability in Guangzhou by optimizing its asset-light model [3][4] Market Positioning - Pony.ai is the only company authorized to operate autonomous ride-hailing services in all four first-tier cities in China, while WeRide has expanded its operations to 11 countries and over 30 cities [4] - The Abu Dhabi market is seen as an ideal environment for Robotaxi commercialization, with WeRide's daily ride volume nearing half that of traditional taxis [6] Industry Trends - The global Robotaxi market is projected to reach $66.6 billion by 2030, with China expected to account for over half of this market [7] - Major players like Waymo and Tesla are also in the Robotaxi space, with Waymo operating over 1500 vehicles and achieving significant ride volumes [7][8] Financial Performance - In Q3 2025, Pony.ai reported total revenue of 181 million yuan, a 72% year-over-year increase, while WeRide achieved 171 million yuan with a 144% growth rate [9] - Both companies are still in the investment phase, with Pony.ai's net loss increasing to 616 million yuan and WeRide's loss narrowing by 70.5% to 307 million yuan [9] - The companies have demonstrated the potential for positive marginal contributions from their Robotaxi businesses, but achieving overall profitability remains a challenge [9] Capital Market Activity - Both Pony.ai and WeRide went public in November 2025, establishing a dual listing in Hong Kong and the U.S. [10] - Pony.ai anticipates needing to deploy tens of thousands of Robotaxi vehicles to reach overall break-even, projecting a fleet size of 100,000 by 2030 [10]

Investment Rating - The report maintains a "Buy" rating for key companies in the autonomous driving sector, including Horizon Robotics, Hesai Technology, Pony.ai, WeRide, Xiaopeng Motors, and Suda Technology [6][8]. Core Insights - The U.S. House of Representatives is reviewing the "SELF DRIVE Act of 2026," which proposes to increase the exemption limit for manufacturers from 2,500 vehicles to 90,000 vehicles, significantly enhancing the potential for Robotaxi mass production [1][2]. - The legislation aims to resolve core obstacles to the scaling of Robotaxi deployment, including the introduction of a "deemed approval" mechanism for exemption applications and establishing federal regulations as a priority over state laws [2][3]. - 2026 is projected to be a pivotal year for the global L4 autonomous driving industry, with the U.S. federal legislative framework becoming clearer and China accelerating its L3 pilot projects and L4 commercialization [1][4]. Summary by Sections Legislative Developments - The "SELF DRIVE Act of 2026" is expected to break a decade-long legislative deadlock, with bipartisan support leading to a clearer timeline for implementation [3]. - Key milestones include committee markup meetings in Q1 2026 and potential integration with the 2026 Surface Transportation Reauthorization Act [3]. Market Expansion - The U.S. Robotaxi market is transitioning from technology validation to large-scale expansion, with companies like Waymo and Tesla ramping up operations [4]. - Waymo's weekly paid orders reached 450,000 by December 2025, with plans to expand operations from 10 to 30 cities [4]. - Tesla's Cybercab is expected to begin mass production in April 2026, further contributing to market growth [4]. Investment Recommendations - The report emphasizes the synchronized development of L4 autonomous driving in both the U.S. and China, suggesting that this convergence will drive significant industry growth [5]. - Companies such as WeRide and Pony.ai are highlighted for their international expansion efforts, while other L4 applications like Robovan and Robotruck are also gaining traction [5].

自动驾驶之心行业交流群来了，关注头部新势力、Tier1、主机厂最新动态、L4赛道融资、技术进展、智驾落 地、行业动态等方向~ 添加小助理微信AIDriver005，备注：昵称+机构/学校+进群。 ...

Core Viewpoint - Citigroup has set a target price of HKD 39.6 for WeRide (00800) and rated it as "Buy" based on several factors [1] Group 1: Investment Outlook - Citigroup has initiated a 90-day upward catalyst observation for WeRide, considering potential developments in the autonomous driving sector [1] - The potential financing round for Waymo in the first quarter of this year implies a valuation of 280 times the sales for 2025, which could lead to a reassessment of WeRide's valuation [1] - Investor confidence is increasing regarding WeRide's possible entry into the Hong Kong Stock Connect (southbound) in June this year [1] Group 2: Industry Dynamics - The competitive landscape in the technology sector between China and the U.S. may prompt China to adopt more proactive policies regarding ADAS (Advanced Driver Assistance Systems) and autonomous taxi services [1] - The light asset business of WeRide is showing positive development [1]

Core Viewpoint - Citigroup has set a target price of HKD 39.6 for WeRide (00800) and rated it as "Buy" based on several factors including potential financing rounds and market confidence [1] Group 1: Valuation and Market Sentiment - Citigroup initiated a 90-day upward catalyst observation for WeRide, considering the potential financing round by Waymo in Q1 of this year, which implies a valuation of 280 times the 2025 sales ratio [1] - There is an increased investor confidence regarding WeRide's potential entry into the Hong Kong Stock Connect (southbound) in June this year [1] Group 2: Industry Dynamics - The competition in the technology sector between China and the U.S. may lead to more proactive policies in China regarding ADAS (Advanced Driver Assistance Systems) and autonomous taxi services [1] - The light asset business of WeRide is developing well, indicating positive growth prospects [1]

Group 1 - Citi has initiated a 90-day upward catalyst observation for WeRide, considering factors such as Waymo's potential financing round in Q1 this year, which implies a valuation of 280 times the 2025 sales ratio, possibly leading to a reassessment of WeRide's valuation [1] - Investor confidence in WeRide potentially entering the Hong Kong Stock Connect (southbound) in June this year has increased [1] - The competition in the technology sector between China and the US may prompt China to implement more proactive ADAS/autonomous driving rental policies [1] Group 2 - The light asset business of WeRide is developing well [1] - Citi has set a target price of HKD 39.6 for WeRide, with a rating of "Buy/High Risk" [1]

Summary of Key Points from the Document Industry Involved - The document discusses the **autonomous driving industry** in the United States, specifically focusing on the implications of new legislation for self-driving vehicles. Core Insights and Arguments - The **SELF DRIVE Act** hearing held by the U.S. House of Representatives on January 13 has received bipartisan support for significantly increasing the exemption cap for autonomous vehicles to **90,000 units per year** [1] - This legislative progress is viewed as a critical step in removing regulatory barriers for the large-scale commercial deployment of **Robotaxi services** [1] - The new regulations are expected to directly accelerate the production process of vehicles without steering wheels, such as **Tesla's Cybercab**, and significantly shorten the timeline for commercialization across the industry [1] Other Important but Potentially Overlooked Content - The establishment of **federal priority** in the legislation indicates a shift towards a more unified regulatory framework for autonomous vehicles, which could enhance investor confidence and attract more capital into the sector [1]

Summary of Conference Call on Robotaxi Developments Industry Overview - The conference discusses the developments in the Robotaxi industry, focusing on key players such as Waymo, Tesla, and Nvidia, along with their respective technologies and market strategies [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26]. Key Players and Their Developments Waymo - Waymo is currently the largest Robotaxi operator globally with a fleet of 2,500 vehicles, although this number is significantly lower than expected [2]. - The company excels in software application, response speed, and supply matching, providing a comprehensive user experience [2]. - Waymo's system is based on rules and high-definition maps, which limits its scalability outside designated areas [1][5]. - The transition to an end-to-end model poses challenges, including regulatory pressures and the complexity of changing its existing technology stack [10]. Tesla - Tesla's Robotaxi does not rely on high-definition maps but uses open-source map data, allowing it to cover more routes and provide a more complete end-user experience [4][5]. - Currently, Tesla operates a limited number of vehicles (150) in Texas and has begun testing fully autonomous operations [4][11][12]. - The cost of Tesla's Robotaxi service is significantly lower than competitors like Uber, with fares from San Francisco to Nvidia headquarters costing under $30 compared to Uber's $50-$60 [4]. - Tesla faces challenges with software stability and low failure rates, which are critical for the success of its Robotaxi operations [13][14]. Nvidia - Nvidia showcased an end-to-end autonomous driving model using the Mercedes CLA, which exceeded expectations during testing [9]. - The company plans to cover all of California by Q1 2026 and gradually expand across North America, although it has decided not to enter the Chinese market for autonomous driving [3][9][23]. - Nvidia continues to offer lidar technology options to clients but has not released a formal Robotaxi solution [3][20]. Competitive Landscape - Other notable competitors in the North American market include Amazon's Zoox, which, despite being a significant player, is lagging in progress compared to Waymo and Tesla [6]. - The performance of competitors like Lucid and Pony.ai is also mentioned, with Waymo being favored due to its strong AI integration and operational experience [8]. Regulatory and Market Challenges - The regulatory environment in the U.S. and China is described as aggressive, with both countries making significant strides in autonomous driving regulations [3][26]. - Local government support varies, with some regions in China showing superficial support for Robotaxi initiatives, while the U.S. faces challenges due to the autonomy of individual states [24]. User Experience and Technology Differences - Waymo offers a more polished user experience, including features like music integration and user onboarding, while Tesla leverages its existing ecosystem for a familiar experience [15]. - Differences in remote takeover capabilities between Waymo and Tesla are noted, with Waymo allowing remote monitoring and control of vehicles [16]. Conclusion - The Robotaxi industry is rapidly evolving, with key players like Waymo and Tesla leading the charge. However, challenges related to scalability, regulatory compliance, and technology integration remain significant hurdles for all companies involved in this space [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26].

点击下方 卡片 ，关注" 自动驾驶之心 "公众号 戳我-> 领取 自动驾驶近30个 方向 学习 路线 很荣幸自动驾驶之心邀请到本文作者 刘彦淏，为大家分享这篇工作。今晚七点半，锁定自动驾驶之心直播间~ 论文标题 ： DriveVGGT: Visual Geometry Transformer for Autonomous Driving 论文链接 ： https://arxiv.org/abs/2511.22264 分享介绍 >>直播和内容获取转到 → 自动驾驶之心知识星球 点击按钮预约直播 前馈重建技术近年来备受关注，其中视觉几何Transformer（VGGT）是典型代表。然而，由于VGGT的设计初衷与自动驾 驶任务的先验知识存在本质差异，将其直接应用于自动驾驶（AD）系统会导致次优结果。在自动驾驶场景中，需重点考 量三类关键新先验：（i） 相机视图重叠度极低 ——自动驾驶传感器配置的核心目标是以低成本实现360度全场景覆盖； （ii） 相机内参与外参已知 ——这不仅为输出结果提供了更多约束，更使得绝对尺度估计成为可能；（iii） 相对位置固 定 ——尽管自车处于运动状态，所有车载相机的相对位置始终保持不 ...