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 ...

Core Viewpoint - The "Mosu Zhixing" action plan aims to accelerate the transformation of intelligent connected technology innovation into industrial competitiveness, promoting the commercialization of high-level autonomous driving in Shanghai [1][3]. Group 1: Action Plan Overview - On January 7, 2023, three departments in Shanghai jointly issued the "Mosu Zhixing" action plan to seize opportunities in automotive intelligence development and foster a new ecosystem for intelligent connected vehicles [1]. - The plan emphasizes a model-driven approach, application demonstration, industrial collaboration, and policy support to advance the construction of the leading area for high-level autonomous driving [1][3]. Group 2: 2027 Goals - By 2027, China aims to establish a globally leading high-level autonomous driving area, with L4 technology implemented in rental and heavy truck scenarios, achieving over 6 million passenger trips and over 800,000 TEU in cargo [2]. - The plan includes the establishment of a digital twin training platform, with 2,000 square kilometers of open testing areas and over 5,000 kilometers of roads covering various scenarios [2]. Group 3: Diverse Application Scenarios - The action plan promotes the simultaneous advancement of passenger vehicles, commercial vehicles, and unmanned equipment [2]. - In the passenger vehicle sector, there will be organized intelligent taxi demonstration operations and pilot projects for L3 vehicles, while commercial vehicles will focus on technology applications in key urban areas and transport hubs [2]. Group 4: Innovation Ecosystem and Policy Support - The plan calls for the Shanghai Economic and Information Commission to advance key technology breakthroughs, cultivate quality enterprises, and promote collaboration in intelligent driving model research [3]. - It emphasizes the need for comprehensive support through policies, finance, talent, and regional collaboration to facilitate the development of the intelligent connected vehicle industry [3].

Group 1 - Trump's threat to impose tariffs on Denmark and other European countries has led to significant market volatility, with European stock markets declining and a surge in demand for safe-haven assets like gold and silver [2][7] - The China Securities Regulatory Commission (CSRC) has outlined key tasks for 2026, focusing on market stability, regulatory enforcement, and promoting the development of listed companies [2][7] - Five leading solar companies, including Tongwei Co. and Longi Green Energy, have announced a combined expected loss exceeding 28.9 billion yuan due to industry challenges such as supply-demand imbalances and rising raw material costs [2][7] Group 2 - Tesla's CEO Elon Musk has announced the restart of the Dojo 3 project, with the new AI5 chip expected to have five times the computing power of the current HW4 chip, impacting the rollout of full self-driving capabilities [3][8] - Rare earth prices have been rising, with a projected supply-demand gap of 140,000 tons by 2030, driven by strong demand from the global electric vehicle sector [3][8] - Several small and medium-sized banks have raised deposit rates as part of a strategy to attract deposits amid low net interest margins, although future rates may stabilize or slightly decrease [3][9] Group 3 - Rongbai Technology is under investigation by the CSRC for misleading statements regarding a significant contract, raising concerns about its ability to fulfill orders due to production capacity issues [4][9] - The minimum margin requirement for financing purchases on the Shanghai and Shenzhen stock exchanges has been increased from 80% to 100% for new contracts, aimed at controlling market leverage risks [4][9] - The 2025 Hurun Report has ranked Cambrian as the top AI company in China, valued at 630 billion yuan, with an increasing number of AI chip companies listed, reflecting a shift towards domestic computing power independence [5][10]

Investment Rating - The report rates the computer industry as "Outperform the Market" [2] Core Insights - The commercialization of AI is accelerating, with significant developments in intelligent agents and video generation technologies [2][10][12] - Ant Group and Google have launched a Universal Commercial Protocol (UCP) to standardize AI-driven commercial interactions, enhancing seamless collaboration across various systems [10][11] - Kuaishou's Keling AI has achieved a monthly revenue of over $20 million (approximately 140 million RMB) as of December 2025, indicating rapid commercialization in the video generation sector [12][13] - Alibaba's Qianwen App has integrated deeply with its ecosystem, transitioning AI capabilities from simple chat functions to executing complex tasks, marking a new era in AI applications [15][16] - Shanghai's "Mosu Zhixing" initiative aims to scale L4 autonomous driving applications by 2027, establishing a leading position in the global smart connected vehicle industry [19][20] - The U.S. has threatened storage chip manufacturers with a 100% tariff unless they increase domestic production, highlighting the geopolitical dynamics in the semiconductor industry [22][23] Summary by Sections AI Commercialization - Ant Group and Google have introduced UCP, a new open standard for intelligent agents that facilitates seamless commercial interactions across various platforms [10][11] - Kuaishou's Keling AI has seen a significant increase in revenue, reaching an annual run rate of $240 million (approximately 1.68 billion RMB) by December 2025, driven by enhanced product capabilities and computational power [12][13][14] Integration of AI in Ecosystems - Alibaba's Qianwen App has integrated with major services like Taobao and Alipay, enabling it to perform real-world tasks such as ordering food and booking travel, thus evolving into a comprehensive AI assistant [15][16][17] Autonomous Driving Initiatives - Shanghai's "Mosu Zhixing" plan aims for large-scale deployment of L4 autonomous driving technology by 2027, with specific targets for passenger and freight transport [19][20][21] Semiconductor Industry Dynamics - The U.S. Commerce Secretary has warned storage chip manufacturers of potential tariffs, emphasizing the need for increased domestic production and the strategic importance of semiconductor independence [22][23]

Group 1 - Chery is actively benchmarking Tesla's Full Self-Driving (FSD) system, aiming not only to match but to surpass it [2] - The company is sending personnel to the U.S. to experience Tesla's FSD and Grok model combination, identifying gaps to accelerate its progress [2] - Autonomous driving is becoming a core competitive advantage for automakers, serving as a key component of technological barriers and a driver for business model upgrades [2] Group 2 - The SAE defines six levels of autonomous driving from L0 to L5, with L5 representing full automation where the system can handle all driving tasks without human intervention [4] - Level 2 advanced driver assistance systems (ADAS) have become mainstream, with penetration rates in China's passenger car market rising from 23.5% in 2021 to 42.4% in the first half of 2023 [6] - The Ministry of Industry and Information Technology in China has granted the first L3 conditional autonomous driving vehicle licenses, marking a significant step towards clearer responsibilities and real-world applications [8] Group 3 - NVIDIA's CEO predicts that in the next decade, a significant portion of vehicles will be autonomous or highly autonomous, potentially reaching a scale of one billion vehicles, all powered by AI [8]

Core Viewpoint - Baidu Group-SW (9888.HK) experienced a notable stock price increase, reaching a high of 150.3 HKD, marking the highest level since August 2023, before closing at 147.4 HKD, up 1.24% [1] Group 1 - Baidu's subsidiary, Luobo Kuaipao, has officially launched a fully autonomous driving service for the public in Abu Dhabi in collaboration with UAE-based AutoGo [1] - This marks the first time Luobo Kuaipao has introduced a public-facing fully autonomous driving service overseas [1]

一般对应闭环仿真或场景重建的算法岗位，日常工作聚焦在： 一般公司需要 5-20 人的算法团队支撑闭环仿真中重建的优化。除此之外，在和其他小伙伴的交流过程中，我们发现 云端的数据生产 也有需求，比如BEV视角下 的静态路面重建，即2DGS重建，可以应用到静态真值生产中，比如RoGS。最近小米的ParkGaussian把GS应用到 泊车场景中 。综合来说，每个方向都需要至少10 人左右的算法团队规模支撑最基本的功能需求。 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 上周有企业做招聘的朋友和柱哥聊了聊，26年需要在重建方向投入一些HC，借这个机会，和大家盘一下GS在自驾业内的职位和需求。 对于重建来说，主要是用于测试的闭环仿真。闭环仿真的用途比较明确，对于一个离线的clip数据，用3DGS重建动静态元素，验证新模型在重建clip上的效果，是 否可以预测合理的新轨迹，沿用新的轨迹能否正常行驶。 添加助理领取优惠！ 讲师介绍 Chris：QS20 硕士，现任某Tier1厂算法专家，目前从事端到端仿真、多模态大模型、世界模型等前沿算法的预研和量产，参与过全球TOP ...

Core Viewpoint - Baidu Group's stock price increased against market trends, reaching a new high since August 2023, driven by the launch of its fully autonomous driving service in Abu Dhabi [1] Group 1: Stock Performance - Baidu's stock rose by 3.23% during trading, reaching 150.3 HKD, before closing up 1.24% at 147.4 HKD [1] Group 2: Business Development - Baidu's subsidiary, Luobo Kuaipao, partnered with UAE's AutoGo to officially launch a fully autonomous driving commercial operation for the public in Abu Dhabi [1] - This marks Luobo Kuaipao's first overseas launch of a public-facing fully autonomous driving service [1]

Core Insights - The article discusses the development of SparseOccVLA, a new Vision-Language-Action model that effectively bridges the gap between Vision Language Models (VLMs) and Semantic Occupancy, addressing challenges in autonomous driving scenarios [2][3][32] Group 1: Model Development - SparseOccVLA utilizes a lightweight Sparse Occupancy Encoder to generate compact yet information-rich sparse occupancy queries, serving as the sole bridge between visual and language inputs [3][14] - The model integrates a language model-guided Anchor-Diffusion planner, which features decoupled anchor scoring and denoising processes, significantly enhancing planning performance and stability [3][20] Group 2: Performance Metrics - SparseOccVLA demonstrates superior performance in various benchmarks, achieving a 7% relative improvement in the CIDEr metric on the OmniDrive-nuScenes dataset compared to the current best methods [3][23] - In the Occ3D-nuScenes dataset, SparseOccVLA also surpasses state-of-the-art performance in future occupancy prediction [24] Group 3: Technical Challenges - Traditional VLMs face issues such as token explosion and limited spatiotemporal reasoning capabilities, while Semantic Occupancy models struggle with dense representations that are difficult to integrate with VLMs [4][9] - The article highlights the limitations of existing methods in effectively combining VLMs and occupancy models, which have developed independently in the autonomous driving field [4][11] Group 4: Experimental Results - The experimental results indicate that SparseOccVLA requires significantly fewer tokens (as low as 300) to achieve competitive performance compared to methods that require over 2500 tokens, ensuring efficient inference [23] - The model's ability to recognize both tangible objects and non-geometric elements, such as traffic lights and lane markings, is attributed to its end-to-end design that retains visual signals from the original images [31]