Core Insights - WeRide (文远知行) is a leading autonomous driving technology company, recognized for its innovative solutions and significant market presence [1][2]. Financial Performance - As of March 31, 2025, WeRide reported total revenue of 72.437 million RMB, reflecting a year-on-year growth of 1.77% [1]. - The company experienced a net loss attributable to shareholders of 385 million RMB, which represents a year-on-year increase of 17.73% [1]. Upcoming Events - WeRide is scheduled to disclose its mid-year financial report for the fiscal year 2025 on August 20, 2023 [1]. Company Overview - WeRide, established in 2017 and registered in the Cayman Islands, operates through its domestic entity and has conducted autonomous driving research, testing, and operations in 30 cities across 7 countries [1]. - The company has been operational for over 1,600 days and holds autonomous driving licenses in China, the USA, the UAE, and Singapore, making it a unique player in the industry [1]. Product and Service Offerings - WeRide has developed a comprehensive product matrix that includes Robotaxi, Robobus, Robovan, Robosweeper, and Advanced Driving Solutions, catering to various sectors such as smart mobility, smart logistics, and smart sanitation [2]. - The company is recognized for its commercial revenue leadership among global peers in the autonomous driving sector [2]. Strategic Partnerships - WeRide has established strategic partnerships with several top global manufacturers and suppliers, including the Renault-Nissan-Mitsubishi Alliance, Yutong Group, GAC Group, and Bosch [2]. Industry Recognition - In 2023, WeRide was ranked eighth on Fortune's list of companies changing the world, making it the only Chinese company to enter the top ten [2].

今天自动驾驶之心很荣幸邀请到BezierGS工作的一作马梓培，为大家分享这篇ICCV'25中稿的新工作！今晚七点 半，自动驾驶之心直播间不见不散~ 1. 构建一个高质量街景世界，供自驾模型在其中训练、探索，减少数据采集的成本； 2. 减少对bounding box精确性的依赖，目前业界以及开源自驾数据集采集的准确性不是很高，bounding box的 标注不精确； 3. 这篇是对自驾世界的学习与探索，未来会探索一个真正的自驾世界模型，该工作只能实现轨迹内插，无法轨 迹外插。 论文链接：https://arxiv.org/abs/2506.22099 代码代码：https://github.com/fudan-zvg/BezierGS 复旦大学ICCV 2025中稿的最新工作！自动驾驶场景的真实重建对于开发闭环仿真至关重要。大多数现有方法依 赖于目标的位姿标注，使用这些位姿来重建动态目标并在渲染过程中实现动态重建。这种对高精度目标标注的依 赖限制了大规模和广泛场景的重建。为了解决这一挑战，复旦大学的团队提出了Bezier curve Gaussian splatting （BezierGS），该方法使用可学习的 ...

写在前面 & 笔者的个人理解 扩散模型在自动驾驶场景视频生成中已经展现出比较有前景的视觉生成质量。然而，现有的基于视频扩散的世界模型在灵活长度、长时序预测以及轨迹规划方面 仍存在不足。这是因为传统视频扩散模型依赖于对固定长度帧序列的全局联合分布建模，而非逐步构建每个时间下的局部化分布。本研究提出 Epona ，一种自回 归扩散世界模型，通过两项关键创新实现局部时空分布建模：1） 解耦的时空分解 ，将时间动态建模与细粒度未来世界生成分离；2） 模块化的轨迹与视频预测 ，通过端到端框架无缝整合运动规划与视觉建模。本文的架构通过引入一种新的"链式前向训练策略"（chain-of-forward training strategy），在实现高分辨率、长持 续时间生成的同时解决了自回归循环中的误差累积问题。实验结果表明，与现有方法相比，Epona在FVD指标上提升7.4%，预测时长可达数分钟。该世界模型进一 步可作为实时端到端规划器，在NAVSIM基准测试中优于现有端到端规划器。 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15个 方向 学习 路线 今天自动驾驶之心为大家分享 地平线联合 ...

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15个 方向 学习 路线 亲身经历：滴滴自动驾驶感知算法岗三轮技术面复盘！ 最近参加了滴滴自动驾驶部门感知算法工程师的面试，总共经历了三轮技术面，每一轮的面试官都问得非常深 入，尤其聚焦在项目细节和技术原理上。这里必须提醒大家： 面试前务必吃透简历上的每一个字！ 一旦被问 住，场面真的会很尴尬。 Q4：了解Anchor-Free检测吗？简述FCOS的核心流程。 重点考察对Anchor-Free代表算法FCOS的理解。 Q5：是否接触过端到端（End-to-End）检测算法？ 滴滴在自动驾驶领域布局多年，技术积累相当深厚。他们和广汽埃安联合成立的"广州安滴科技"专注于L4级无 人驾驶研发，投入力度很大。今年想冲击自动驾驶方向的同学，滴滴绝对是一个值得重点关注的选项。 第一轮技术面试回顾： Q1：自我介绍 面试官主要围绕我研究生期间的科研产出和研究方向进行了针对性提问。 Q2：论文核心创新点阐述？ 在讨论创新点时，面试官自然地延伸出了一些2D目标检测领域的相关问题进行探讨。 Q3：概述2D目标检测算法的主要演进脉络？ 需要梳理从传统方法到 ...

Domestic News - The Ministry of Industry and Information Technology (MIIT) is soliciting public opinions on three mandatory national standards, including the speedometer for automobiles and motorcycles, with a deadline of August 30, 2025. The revisions include changes to the scale value marking and requirements for electronic digital speedometers [2][3] - Starting January 1, 2026, new regulations will require all newly approved passenger vehicles to be equipped with Anti-lock Braking System (ABS) to enhance vehicle stability during emergency braking. Additionally, the new standards will prohibit the default "single pedal mode" for energy recovery systems in electric vehicles [4][5] - In the first five months of 2025, Shaanxi Province exported 155,000 new energy vehicles, marking a 173% increase year-on-year, surpassing the total export volume of 147,000 vehicles for the entire year of 2024. The province's total vehicle production reached 789,000 units, a 29.3% increase [6][7] - CATL and Geely have signed a strategic cooperation agreement to deepen collaboration in battery technology and supply chain development [8] - Changan Automobile has opened its first flagship store in Norway, launching the Deep Blue S07 model, which is tailored for the European market [9] - BYD Energy has signed a cooperation framework agreement with State Grid Hunan Comprehensive Energy Service Co., aiming for deep collaboration in distributed energy storage [10][11] - Pony.ai has initiated autonomous driving pilot tests in Dubai, with plans for commercial operation by 2026 [12] - Faurecia China and GAC Huawang have established a joint venture to enhance their competitive edge in the market [13] International News - In June, the UK saw a 6.8% year-on-year increase in new car registrations, with electric vehicle demand rising by 39.1% [14] - The Czech Republic has become the second European country to legally allow Level 3 autonomous vehicles on public roads, effective January 1, 2026 [15] - Hyundai has begun construction of a hydrogen fuel cell factory in Ulsan, South Korea, expected to be operational by 2028 [16] - Tesla has expanded its Full Self-Driving (FSD) testing to Spain, following previous expansions in other European countries [17] Commercial Vehicles - Sichuan Province is promoting the development of hydrogen fuel cells and hydrogen-powered heavy trucks as part of its new industrial strategy [18] - Dongfeng Motor has delivered its first batch of the new Dongfeng Dolika Kangluwei light trucks [19] - BYD's Shark pickup has become the best-selling domestic pickup in Australia, with a 368% increase in registrations year-on-year [20] - Guangzhou Yuanchen New Energy Commercial Vehicle Company has officially launched, focusing on the electrification of public transport and logistics vehicles [21]

Core Viewpoints - The future will require humans to take responsibility, as AI cannot assume accountability for its actions [2][35] - Kevin Kelly predicts that by 2049, smart glasses will replace smartphones, with a strong belief that Chinese companies will lead this innovation [2][70] - The book "2049: The Possibilities of the Next 10,000 Days" aims to inspire imagination about a better future rather than making definitive predictions [8][10] Group 1: Future of Technology - AI will not replace jobs but will change the nature of work, requiring humans to adapt to new roles [28][29] - The emergence of smart glasses faces significant technological challenges, including power supply and field of view, which may take 5 to 25 years to overcome [59][63] - The concept of a "mirror world" will emerge, where AI and smart glasses create a digital twin of the real world, enhancing interaction and experience [47][49] Group 2: Education and Skills - Future education should focus on teaching students how to learn effectively, ask good questions, and manage their own learning pace [25][27] - The ability to ask insightful questions will become more valuable than simply finding answers, as AI can provide information readily [27][29] Group 3: Media and Content Creation - The role of journalists will remain crucial, as they will be responsible for verifying information generated by AI [33][34] - YouTube is expected to become a dominant platform for media, especially as immersive technologies develop [43][45] Group 4: Robotics and Automation - Human-like robots are likely to be used in industrial settings before entering households, due to the complexity of home environments [80][81] - The widespread adoption of humanoid robots in homes is not expected within the next 25 years [81][87] Group 5: Space Exploration - The idea of colonizing Mars is viewed skeptically, with a belief that the primary value lies in technological advancements rather than establishing cities [97][101] - Space exploration will focus more on learning and technology development rather than direct economic benefits [99][100] Group 6: Future of Silicon Valley - New startups are expected to emerge, potentially changing the landscape of Silicon Valley, but the fundamental dynamics of the ecosystem may remain [126][130] - AI could transform the way startups are funded and developed, possibly leading to more crowdfunding models [131]

忘掉科技"七巨头"吧！今年上半年，美股投资者正学着爱上"未盈利的858"…… 有迹象显示，网红股(Meme)和那些迄今仍未能实现盈利的公司，在过去几个月正重获投资者 的青睐，并支撑着推动近期市场创下新高的反弹行情。 根据Bespoke投资集团的分析，在罗素3000指数成份股中，自4月8日以来涨幅超过200%的 总计有14家公司，而其中有多达10家未能报告实现任何利润。同时，截至6月下旬，罗素 3000指数中合计858家未能实现业绩盈利的股票的平均涨幅，达到了惊人的36%，表现远优 于那些实现盈利的公司。 从Avis Budget Group、Carvana到Aeva Technologies，这些新一批网红股的大涨，似乎表 明投资者正重新燃起了投机热情。 这也让一些分析师联想到了2021年的"散户大战华尔街"时 期， 当时近乎于零利率和新冠疫情的政府刺激措施，推动了游戏驿站、AMC等具有高度投机 性股票的疯狂上涨。 盈透证券首席策略师Steve Sosnick表示："我们还没有看到全面的'冲向垃圾股'(flight-to- crap)浪潮，但很明显，推动许多此类股票活动背后的动机，并非是基于贴现现金流的严谨考 ...

Domestic News - The National Development and Reform Commission of China signed cooperation documents with Brazilian authorities, including a memorandum on the second phase of strategic alignment and cooperation in artificial intelligence [2] - The State Administration for Market Regulation exposed six typical cases of "involution" competition in product quality, including a case involving Shaoxing Hongguan New Energy Co., which sold substandard "liquid wax alcohol ether composite diesel fuel" [3] - In the first five months of 2025, the cumulative transaction volume of second-hand cars in China reached 7.91 million units, a year-on-year increase of 0.6%, while the transaction value was 516.4 billion yuan, a year-on-year decrease of 2.1% [4] International News - Tesla's Cybertruck sales are sluggish, with a global delivery of approximately 384,000 units in Q2, a year-on-year decline of 13.5%, marking the largest drop in the company's history. Deliveries of "other models" (including Model S, Model X, and Cybertruck) were about 10,400 units, down 52% year-on-year. Cybertruck sales in Q1 were estimated to be between 5,000 to 6,000 units, far below the target of 250,000 units per year set by Elon Musk for 2025 [5] - Japanese Prime Minister Kishida stated that Japan is prepared for all possible tariff scenarios and will defend its interests while promoting zero tariffs on automobiles [6] - Dubai's Roads and Transport Authority signed a memorandum with Pony.ai to launch pilot tests for autonomous vehicles, with plans for phased implementation starting in 2025 and full commercial operation by 2026 [7] - The EU is experiencing significant internal divisions regarding trade negotiations with the US, with countries like Germany and Italy pushing for quick agreements, while France, Spain, and Denmark express concerns about potential concessions. The outlook for negotiations is not optimistic [8] Corporate News - NIO's CEO Li Bin announced that the starting price for the L90 model will be below 300,000 yuan, with display vehicles set to arrive on July 10 and pre-sales commencing [11] - Li Xiang, CEO of Li Auto, emphasized the importance of open communication among new energy vehicle leaders and suggested that Xiaomi's success in the automotive sector requires a full commitment [12] - BYD announced a strategic partnership with Shanghai LEGO Land Resort to focus on green travel and immersive driving experiences for children aged 2-12 [13] - TSMC is reportedly delaying the construction of its second factory in Japan due to potential tariffs from the Trump administration, as the company accelerates investments in the US [14] - New Stone Technology's autonomous vehicles signed a strategic cooperation agreement with the Shenzhen Car Rental Industry Association, aiming to provide 1,000 autonomous vehicles to enhance smart logistics [15]

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15个 方向 学习 路线 当前自动驾驶技术正处于从辅助驾驶（L2/L3）向高阶无人驾驶（L4/L5）跨越的关键阶段。如果你对自动驾驶 有浓厚的兴趣，并且想找业内最专业的大佬交流，那么这个圈子一定没错！技术迭代的浪潮下暗藏职业焦虑。 对职场老人而言 ，传统以激光雷达为核心的感知算法工程师可能面临路线冲击——特斯拉的纯视觉方案依托成 本优势和算法革新，正在撼动多传感器融合的主流地位；而规划控制领域从PID到强化学习的转型，也让依赖传 统控制理论的从业者陷入技能升级的紧迫感。 学生新手则陷入"选择困难症" ：感知算法赛道因头部企业技术垄 断加剧内卷，数据闭环工程师需要同时掌握分布式计算与AI模型调优能力，而新兴的车路协同方向又要求跨界 融合通信与交通系统工程知识。当禾赛科技将激光雷达成本降至200美元、比亚迪宣布自研体系内价格再降70% 时， 技术红利背后实则是从业者必须持续奔跑的生存法则，这种技术路线的不确定性与知识体系的重构压力， 正在重塑自动驾驶人才市场的竞争格局。 后处理，写逻辑建议转行业可以，不要换方向，gap还是蛮大。现在很多人 ...

Group 1 - Tesla's Robotaxi service in Austin, Texas experienced an incident where a fully autonomous Model Y collided with a parked Toyota vehicle, fortunately resulting in no serious injuries or significant vehicle damage [1] - A video captured by YouTuber DirtyTesla shows the Model Y accelerating unexpectedly while exiting a dimly lit alley, leading to the collision with the Toyota [3] - Other users of the Robotaxi service reported additional issues, including sudden emergency stops and incorrect driving on the wrong side of the road [3] Group 2 - Tesla CEO Elon Musk advocates for a vision-based approach to autonomous driving, emphasizing the superiority of using cameras and neural networks over expensive lidar technology [3] - In contrast, competitors like Waymo, which utilize a combination of cameras, lidar, and radar, have also faced safety incidents, including multiple recalls due to collisions with obstacles [3] - The autonomous driving industry faces significant challenges in improving technology stability and ensuring passenger safety, as highlighted by the recent incident involving Tesla's Robotaxi [4]