编辑丨 机器之心 点击下方 卡片 ，关注" 具身智能之心 "公众号 >> 点击进入→ 具身 智能之心 技术交流群 更多干货，欢迎加入国内首个具身智能全栈学习社区： 具身智能之心知识星球（戳我） ，这里包含所有你想要的! 随着 VLA（Vision-Language-Action）模型的兴起，端到端自动驾驶正经历从「模块化」向「大一统」的范式转移。然而，将感知、推理与规划压缩进单一模型 后，主流的自回归（Auto-regressive）生成范式逐渐显露出局限性。现有的自回归模型强制遵循「从左到右」的时序生成逻辑，这与人类驾驶员的思维直觉存在本 质差异 —— 经验丰富的驾驶员在处理复杂路况时，往往采用「以终为始」的策略，即先确立长期的驾驶意图（如切入匝道、避让行人、靠边停靠），再反推当 前的短期操控动作。此外，基于模仿学习的模型容易陷入「平均司机」陷阱，倾向于拟合数据分布的均值，导致策略平庸化，难以在激进博弈与保守避让之间灵 活切换。 针对上述痛点， 复旦大学与引望智能联合提出了 WAM-Diff 框架 。该研究创新性地将 离散掩码扩散模型 （Discrete Masked Diffusion）引入 VLA 自动 ...

Group 1 - The Ministry of Industry and Information Technology has officially announced the first batch of L3-level vehicle access permits, marking a significant step towards the commercialization of L3-level autonomous driving in pilot cities like Chongqing and Beijing [1] - L3-level "conditional autonomous driving" introduces a shift in driving responsibility from human to machine, raising concerns about system reliability, algorithm decision-making, and sensor performance [1] - The emergence of "intelligent driving insurance" products in the market is primarily a safety net for car manufacturers or intelligent driving solution providers, rather than genuine insurance products [1] Group 2 - Heng Rui Medicine announced that its SHR-A1904 injection has been included in the list of breakthrough therapeutic varieties by the National Medical Products Administration, highlighting its potential in the ADC drug development field [2] - The ADC drug market is expected to grow significantly, with Chinese companies emerging as global innovation engines in this sector [2] - The A-share market has shown a continuous upward trend, with the MSCI Emerging Markets Index increasing nearly 30% year-to-date, indicating a favorable investment environment for emerging markets [2] Group 3 - The People's Bank of China continues to release liquidity into the market through medium-term lending facilities (MLF), maintaining a "stable and loose" liquidity management approach [3] - In December, the central bank conducted a 400 billion yuan MLF operation, resulting in a net liquidity injection of 1,888 billion yuan, ensuring stable financial market operations at year-end [3] - The total net MLF injection for the year 2025 is projected to exceed 1 trillion yuan, supporting market liquidity [3] Group 4 - The Asian currency market is experiencing significant divergence, with the Japanese yen and South Korean won facing depreciation pressures, while the Chinese yuan shows a strong rebound [4] - Japan and South Korea are actively implementing measures to stabilize their currencies amid unprecedented depreciation pressures [4] - A new climate-related disclosure guideline has been introduced, focusing on governance, strategy, risk management, and metrics for corporate sustainability [4] Group 5 - The low-altitude economy is recognized as a promising future industry, with significant growth potential and increasing investment from capital markets [5] - The development of the low-altitude economy faces challenges such as business model exploration and infrastructure improvement, which need to be addressed for sustainable growth [6] - The IPO underwriting amount for securities firms in 2025 is expected to nearly double year-on-year, indicating a strong market for capital raising [6] Group 6 - A new platform for integrating and acquiring polysilicon production capacity in the photovoltaic industry has been established, aimed at addressing excessive competition [7] - The photovoltaic industry is facing challenges, including a projected decline in domestic demand and uncertainty regarding the impact of rising silicon material prices on downstream prices [7] - The cost increase of auxiliary materials is likely to delay the profitability of downstream battery and component sectors [7] Group 7 - The capital market has seen active financing, with a significant increase in funds raised through private placements, which has benefited securities firms [8] - The total amount raised through private placements by A-share listed companies has increased by over 375% year-on-year, providing more business opportunities for brokers [8] Group 8 - The People's Bank of China is focusing on maintaining capital market stability through various monetary policy tools, emphasizing the importance of supporting the market [9] - In 2025, 111 companies successfully listed on the A-share market, raising a total of 125.32 billion yuan, with a significant portion from strategic emerging industries [9] - Over 200 major asset restructuring announcements have been made in the A-share market, primarily in key sectors such as semiconductors and information technology [9]

Core Viewpoint - Waymo has temporarily suspended its autonomous taxi service in the San Francisco Bay Area due to severe weather warnings, highlighting operational challenges and safety concerns in adverse conditions [1][4][5]. Group 1: Service Suspension and Weather Impact - Waymo's autonomous taxi service was halted due to a flood warning issued by the National Weather Service, which is in effect until Friday evening [4][5]. - The service interruption follows a recent incident where multiple Waymo autonomous vehicles became immobilized at intersections during a power outage, causing significant traffic disruptions [5]. - Waymo plans to update its fleet to enhance operational reliability during power outages [5]. Group 2: Operational Scope and Future Plans - Currently, Waymo operates commercial autonomous services in five U.S. markets, including Austin, San Francisco Bay Area, Phoenix, Atlanta, and Los Angeles, with plans to reduce this to three by the end of 2024 [5]. - The company aims to significantly expand its service range both domestically and internationally by 2026 [5]. Group 3: Regulatory and Safety Considerations - Jeffrey Tumlin, former CEO of the San Francisco Municipal Transportation Agency, emphasized the need for regulatory bodies to learn from the chaos caused by Waymo vehicles during the power outage [6]. - Tumlin suggested that regulators should establish a phased system for autonomous vehicle companies to scale operations based on specific testing criteria [6]. - He also highlighted the importance of collecting more data from autonomous taxi companies regarding their performance during emergencies like power outages and natural disasters [6].

南方财经见习记者郭莎 广州报道 "老板"正在成为广东的特产。截至9月3日，广东省登记在册经营主体突破2000.19万户，较2024年末增 长5%，经营主体总量稳居全国第一。 按照1.28亿常住人口计算，平均每6个人里，就有一个"老板"。深圳、广州更是包揽了全国城市"老 板"数量的冠亚军。 这一数字，是2017年突破1000万户后的第二次跨越。进一步对总量进行拆解可以观察到，民企、个体工 商户、外企等经营主体构成多元，新设企业聚焦于新兴产业赛道，含金量同样很高，76家企业入围2025 《财富》中国500强排行榜。 数字背后更为深刻的命题是，广东为何吸引到这么多的市场主体？答案或许在于，广东擅长扮演"时间 的朋友"，陪伴企业"相识于微末，相逢于顶峰"。 多元共生的商业雨林 广东的经营主体图谱，首先以其多元与包容著称。做高精尖科研项目可以创业，"拧螺丝""卖肠粉"也能 成为老板。 庞大的民营经济是这片雨林的主体。截至8月末，全省民营经济组织占比高达96.45%，同比增长 6.47%。其中，超过834万户私营企业构筑起产业的中坚力量。除了大企业之外，超过千万户的"小老 板"——个体工商户广泛分布在广东的各行各业，成为 ...

拾象CEO 李广密 今天推出第一篇的文字版本： 访谈：张小珺 x Freda Duan 《站在2025年之外》是我们推出的年终对话系列，由5篇投资人访谈组成，他们分别从中国本土与硅谷 视角出发，共同描绘我们走过的2025年，并对2026年做预测与前瞻。 5位嘉宾分别是：（排列依据播客的发布时间顺序） 金沙江创投主管合伙人 朱啸虎 真格基金管理合伙人 戴雨森 Altimeter Capital合伙人 Freda Duan 红杉中国合伙人 郑庆生 一个多月前，在2025年11月初，Sam Altman上了一档由美国基金Altimeter Capital创始人主持的播客节 目，在主持人连续追问OpenAI如何为1.4万亿美元级别算力与基础设施承诺买单时，Sam称："If you want to sell your shares, I'll find you a buyer. Enough."（"如果你想卖掉你的股份，我可以帮你找到买家。 够了。"） 随后，AI板块整体出现波动，关于AI是否存在泡沫的讨论进一步升温。 我们今天的嘉宾Freda Duan就来自这个名叫Altimeter Capital的基金，她担任合伙 ...

Core Viewpoint - UniLION is a groundbreaking unified autonomous driving framework developed by the University of Hong Kong, Huazhong University of Science and Technology, and Baidu, which effectively addresses computational efficiency issues in processing large-scale point cloud data and multi-view images using linear group RNN technology [2][3]. Group 1: Project Overview - UniLION is designed to efficiently handle large-scale LiDAR point clouds, high-resolution multi-view images, and temporal data without the need for explicit temporal or multi-modal fusion modules, supporting various configurations seamlessly [4][5]. - The framework aims to simplify the design of multi-modal and multi-task autonomous driving systems while maintaining superior performance across core tasks such as 3D perception, prediction, and planning [3][44]. Group 2: Research Background and Challenges - Current autonomous driving systems face challenges in computational efficiency, multi-modal fusion complexity, temporal information processing, and multi-task learning difficulties [5]. - Traditional Transformer models introduce significant computational overhead due to their quadratic complexity in attention mechanisms when processing long sequences [5]. Group 3: Innovations of UniLION - UniLION features a unified 3D backbone network based on linear group RNN, allowing seamless processing of different modalities and temporal information without explicit fusion modules [8]. - The framework utilizes linear computational complexity to convert multi-view images, LiDAR point clouds, and temporal information into tokens for unified integration in 3D space [8]. - UniLION generates a compact unified bird's-eye view (BEV) representation of heterogeneous multi-modal information and time series, serving as shared features for various downstream tasks [8]. Group 4: Performance Results - UniLION demonstrated competitive and state-of-the-art performance on the nuScenes dataset, achieving 74.9% NDS and 72.2% mAP in 3D object detection, 76.2% AMOTA in multi-object tracking, and 72.3% mIoU in BEV map segmentation [20]. - The strongest temporal multi-modal version of UniLION achieved 75.4% NDS and 73.2% mAP in detection tasks, showcasing its advanced capabilities across multiple evaluation tasks [20]. Group 5: Efficiency and Robustness - UniLION significantly reduces computational resource requirements and inference time through its linear computational complexity, making it suitable for deployment in real-world autonomous driving systems [35]. - The framework exhibits strong robustness against sensor misalignment, maintaining performance even under high misalignment levels [32]. Group 6: Future Prospects - Future work includes expanding UniLION to support additional sensor modalities, applying it in real-world autonomous driving systems, and exploring large-scale pre-training to enhance its generalization capabilities [45].

Core Viewpoint - The Hong Kong Transport Department has issued a new pilot license for autonomous vehicles, aiming to facilitate passenger transport from the Hong Kong-Zhuhai-Macao Bridge to the Hong Kong International Airport [1][2] Group 1: Project Details - The approved testing route for autonomous vehicles spans from the "Transfer Parking Lot" at the Hong Kong-Zhuhai-Macao Bridge to the entrance of the restricted area at the airport [1] - During the initial testing phase, each autonomous vehicle will have a backup operator onboard to take control if necessary [1] - All pilot autonomous vehicles must display a designated label from the Hong Kong Transport Department for identification by other road users [1] Group 2: Future Plans and Applications - The Transport Department anticipates transitioning to passenger operations after successful testing, allowing travelers to park at the automated "Transfer Parking Lot" and take autonomous vehicles to the SkyPier for transfers [2] - The department invites interested organizations or companies to submit applications for pilot licenses, which will be evaluated based on various factors including design, operational scope, and compliance with national or international standards [2]

来源：环球网 2025年成为中国汽车产业转型的关键节点：新能源乘用车渗透率首次突破50%大关，标志着市场从"政策驱动"全面转向"市场驱动"；与此同时，工信部正式 发放首批 L3级有条件自动驾驶车型准入许可，宣告自动驾驶从封闭测试迈入商业化应用新阶段。 可以看到，这两大趋势的叠加，不仅重构了汽车产业的竞争格局，更向激光雷达、域控制器、高精地图等核心产业链释放出明确的规模化需求信号。随着技 术成熟度提升与成本下降，自动驾驶相关零部件产业正迎来爆发式增长，成为支撑汽车产业高质量发展的新引擎。 新能源汽车渗透率突破临界点，构筑智能化发展坚实底座 新能源汽车市场的持续扩容，已从量变积累走向质变突破。中国汽车工业协会数据显示，2025年1-11月新能源汽车产销量分别达1490.7万辆、1478万辆，同 比大幅增长31.4%、31.2%，增速远超行业整体水平，已成为驱动我国汽车产业高质量发展的核心增长引擎。 这一数据的突破背后，是消费需求的结构性变革：增换购需求跃居核心驱动力，2025年换购比例预计超60%，Z世代成为消费主力，70%的年轻消费者将智 能驾驶功能列为购车核心考量因素，为自动驾驶技术的规模化应用奠定了坚实的市 ...

Core Viewpoint - The article discusses the distinction between world models and end-to-end models in autonomous driving, clarifying that world models are not a specific technology but rather a category of models with certain capabilities. It emphasizes the trend in the industry towards using world models for closed-loop simulation to address the high costs associated with corner cases in autonomous driving [2]. Course Overview - The course on world models in autonomous driving is structured into six chapters, covering the introduction, background knowledge, discussions on general world models, video generation-based models, OCC-based models, and job-related insights in the industry [5][6][7][8][9]. Chapter Summaries - **Chapter 1: Introduction to World Models** This chapter outlines the relationship between world models and end-to-end autonomous driving, discussing the development history and current applications of world models, as well as various streams such as pure simulation, simulation plus planning, and generating sensor inputs [5]. - **Chapter 2: Background Knowledge** This chapter covers foundational knowledge related to world models, including scene representation, Transformer technology, and BEV perception, which are crucial for understanding subsequent chapters [6]. - **Chapter 3: General World Models** Focuses on popular general world models like Marble from Li Fei-Fei's team and Genie 3 from DeepMind, discussing their core technologies and design philosophies [7]. - **Chapter 4: Video Generation-Based World Models** This chapter delves into video generation algorithms, starting with GAIA-1 & GAIA-2 and extending to recent works like UniScene and OpenDWM, highlighting both classic and cutting-edge advancements in this area [8]. - **Chapter 5: OCC-Based World Models** Concentrates on OCC generation algorithms, discussing three major papers and a practical project, emphasizing the potential for these methods to extend into vehicle trajectory planning [9]. - **Chapter 6: World Model Job Topics** This chapter shares practical insights from the instructor's experience, addressing industry applications, pain points, and interview preparation for positions related to world models [9]. Learning Outcomes - The course aims to provide a comprehensive understanding of world models in autonomous driving, equipping participants with the knowledge to achieve a level comparable to one year of experience as a world model algorithm engineer [10].

自动驾驶之心L4交流群来了，关注L4赛道融资、技术进展、RoboTaxi、RoboBus、RoboVan、无人配送、无人 矿卡、无人重卡等方向~ 添加小助理微信AIDriver005，备注：昵称+机构/学校+进群。 ...