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]

1月19日，百度集团-SW(9888.HK)今日逆势上涨，盘中一度涨3.23%报150.3港元，股价创2023年8月以来新高；最终收涨 1.24%报147.4港元。 消息面上，百度旗下萝卜快跑与阿联酋自动驾驶出行公司AutoGo共同宣布，在阿布扎比正式启动面向公众的全无人驾驶商 业化运营。这是萝卜快跑首次在海外推出面向公众的全无人驾驶出行服务。 港股频道更多独家策划、专家专栏，免费查阅>> 责任编辑：安东 ...

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 ...

百度集团-SW(9888.HK)今日逆势上涨，盘中一度涨3.23%报150.3港元，股价创2023年8月以来新高；最 终收涨1.24%报147.4港元。消息面上，百度旗下萝卜快跑与阿联酋自动驾驶出行公司AutoGo共同宣 布，在阿布扎比正式启动面向公众的全无人驾驶商业化运营。这是萝卜快跑首次在海外推出面向公众的 全无人驾驶出行服务。 ...

自动驾驶要解决的是场景的泛化性，具身要解决的是行为的泛化性。 点击下方 卡片 ，关注" 第一具身范式 "公众号 第一时间获取具身智能 干货 编辑 | 第一具身范式 原文链接： http://xhslink.com/o/9dt9pYOtnY 谷歌waymo最近在一次采访中提到:自动驾驶是最简单的机器人，是最复杂的社交游戏。这启发了我去认真思 考一下自动驾驶和具身机器人的关系。 正好最近抽空把physical intelligence的pi系列论文好好看了下，一直觉得自动驾驶和机器人在技术栈上很类 似，甚至一度觉得自驾其实是机器人的子集，现在看完的想法是两者在量产角度上差异远比想象的大。具身 智能的发展周期可能也和自动驾驶不太一样。 泛化性: 自动驾驶一直想解决的是场景的泛化问题。简单来说，就是自动驾驶其实想要的是对当前场景能够有一个考 虑尽可能全面的理解，然后做出对应的决策。举个具体的例子自动驾驶需要在前面有个锥桶的时候知道刹车 但也需要当前面有个载着锥桶的工程车的时候知道不用刹车。体现的是对场景的认知泛化能力。 比如多段式模块规则时代，对运动信息有预测模块，对语义信息有融合模块，在规划层做场景信息的整合与 理 ...

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 论文作者 | Chenxu Dang等 编辑 | 自动驾驶之心 在自动驾驶领域，视觉-语言模型（Vision Language Models, VLMs） 擅长高层语义理解与推理，而语义占据（Semantic Occupancy） 则能够提供精细、结构化的空间 细节。尽管这两个方向各自取得了显著进展，但目前仍缺乏一种能够有效融合二者的统一方法。 一方面，传统 VLM 在自动驾驶场景中面临 token 数量爆炸 以及 时空推理能力受限 等问题；另一方面，语义占据通过统一且显式的空间表示建模环境，但其表示过于 稠密，难以高效地与 VLM 进行集成。 为了解决上述挑战并弥合 VLM 与占据表示之间的鸿沟， 华科、小米和清华AIR的团队提出了 SparseOccVLA ，一种新的 视觉-语言-动作（Vision-Language-Action, VLA）模型，通过 稀疏占据查询（Sparse Occupancy Queries） 实现了场景理解、占据预测与轨迹规划的 ...

政策与技术双驱 智驾 L3 与 L4 的变局 20260118 摘要 英伟达开源 Alpaca 模型及相关生态，降低 Robotaxi 规模化落地和 L2+技术普及的技术门槛，将行业竞争焦点从算法转向工程和运营能力， 有利于跨区域复制和提升监管效率。 北美自动驾驶政策趋于宽松，简化审批流程、放宽安全要求，旨在促进 技术创新和商业化，为本土及全球车企提供更高效灵活的研发环境，加 速自动驾驶技术进步。 自动驾驶技术的核心短板包括极端天气适配、施工路段识别和用户信任 体系构建。解决方案包括模态增强、视觉与语言理解结合、线控系统冗 余设计以及提供可解释性决策。 国内自动驾驶产业发展需关注技术科学性和长期复现能力，补充可解释 性，重视仿真和物理 AI 应用，优化样本分布。商业化方面，可考虑一次 性放开更大规模 OTA 指标，实施跨城运营。 2026 年中国智能驾驶领域关键进展包括 L3 级别带条件商业化、 Robotaxi 规模扩展（数万台规模）以及自动驾驶企业在模型和方法论 上的迭代优化。 Q&A 英伟达模型开源对 Robotaxi 规模化落地和 L2+级技术普及有哪些核心影响？ 以及对自动驾驶政策中安全可解释性要求 ...

Core Insights - The article discusses the latest developments in the automotive intelligence sector, highlighting advancements in L3 autonomous driving technology and the increasing penetration rates of various intelligent features in vehicles. Group 1: L3 Autonomous Driving Developments - The Ministry of Industry and Information Technology has approved two L3 autonomous driving vehicle models, marking a significant step towards commercial application in China [10]. - Tesla's Full Self-Driving (FSD) technology is expected to receive full approval in China by early 2026, indicating progress in regulatory acceptance [11]. - Xiaopeng Motors has obtained a road testing license for L3 autonomous driving in Guangzhou, furthering its testing capabilities [12]. Group 2: Market Penetration Rates - As of October 2025, the penetration rate of passenger vehicles with L2 and above features reached 33%, a year-on-year increase of 19 percentage points [8]. - The penetration rates for advanced driver-assistance systems (ADAS) such as highway NOA and urban NOA are 33.8% and 16.2%, respectively, with year-on-year increases of 21 and 8 percentage points [8]. - The penetration of 800 million pixel cameras in passenger vehicles has reached 49.7%, up 31% year-on-year [6]. Group 3: Industry Collaborations and Innovations - WeRide's Robotaxi service has successfully launched in over 10 cities globally, demonstrating the commercial viability of autonomous driving technology [13]. - Hiydi Zhijia has become the first company focused on commercial vehicle intelligent driving to be listed on the Hong Kong Stock Exchange, raising approximately 1.422 billion HKD [15]. - RoboSense has secured a contract with Dongfeng Nissan for nearly one million units of digital lidar products, set to begin mass production in 2026 [17]. Group 4: Sensor and Technology Advancements - The penetration rate of laser radar in passenger vehicles has reached 14.3%, with a year-on-year increase of 7.9 percentage points [6]. - The market share of NVIDIA chips in passenger vehicle driving domain controllers has increased to 58%, reflecting a 22.2% year-on-year growth [6]. - The cumulative shipment of Huayang Group's HUD products has surpassed 3.5 million units, solidifying its position as a leading supplier in the global market [16].

Core Viewpoint - Nvidia's launch of the Alpamayo ecosystem in autonomous driving is seen as a significant development, but it is unlikely to disrupt Tesla's FSD dominance due to Nvidia's focus on providing foundational computing power rather than a fully integrated autonomous driving solution [3][4][5]. Group 1: Nvidia's Business Model - Nvidia's business model centers around offering a toolkit for development rather than a plug-and-play autonomous driving system, encouraging clients to leverage their computing power for iterative model development [4][5][6]. - The company aims to reduce the initial investment costs for clients in autonomous driving research, promoting a collaborative ecosystem rather than direct competition with Tesla [6][9]. Group 2: Competitive Landscape - Nvidia does not have a strong incentive to challenge Tesla directly, as Tesla is its largest customer, and Nvidia benefits from a diverse competitive landscape in the autonomous driving sector [6][9]. - The lack of a dominant player like Tesla is seen as beneficial for Nvidia, as it encourages widespread GPU purchases among various automotive companies [9][10]. Group 3: Data and Simulation Challenges - Nvidia's data collection capabilities are limited compared to Tesla's extensive fleet, which hampers its ability to compete effectively in the autonomous driving space [10][11]. - The Physical AI dataset released by Nvidia, while extensive, is primarily focused on the U.S. and Europe, and lacks the breadth needed for comprehensive autonomous driving development [10][11][13]. - Nvidia's reliance on simulation technology for data generation is seen as a potential weakness, as effective simulation requires substantial real-world data to be truly effective [12][14]. Group 4: Market Dynamics - The autonomous driving market has evolved significantly since Google's initial foray in 2009, with the current landscape favoring companies that can deliver practical, scalable solutions rather than just prototypes [15][16]. - Nvidia's collaboration with Mercedes for production-level autonomous driving has faced delays, indicating challenges in achieving competitive market readiness [17]. - In China, the autonomous driving landscape is characterized by intense competition among local manufacturers, which complicates Nvidia's strategy to maintain its ecosystem [18][19].

股票研究/[Table_Date] 2026.01.18 [Table_Industry] 计算机 上海发布"模速智行"行动计划，自动驾 驶产业驶入加速赛道 [Table_Invest] 评级： 增持 | [姓名table_Authors] | 电话 | 邮箱 | 登记编号 | | --- | --- | --- | --- | | 杨林(分析师) | 021-23183969 | yanglin2@gtht.com | S0880525040027 | | 魏宗(分析师) | 021-23180000 | weizong@gtht.com | S0880525040058 | | 吕浦源(分析师) | 021-23183822 | lvpuyuan@gtht.com | S0880525050002 | | 朱瑶(分析师) | 021-23187261 | zhuyao@gtht.com | S0880526010002 | 本报告导读： 行 业 跟 踪 报 告 证 券 研 究 报 告 研 究 请务必阅读正文之后的免责条款部分 股 票 1 月 7 日三部门联合印发《上海高级别自动驾驶引领区"模速智行"行动计 ...