小马智行：一季度Robotaxi收入同比增200%，车队规模年底达千台 金十数据5月20日讯，小马智行发布2025年第一季度财报。财报显示， 一季度总营收为1.02亿元，同比 增长12%。Robotaxi业务收入达1230万元，同比增长200%，其中乘客车费收入同比增长800%。小马智 行正全力推进Robotaxi量产，2025年底车队规模将达到千台。 ...

小马智行第一财季营收1,400万美元，同比增长12%。 第一财季毛利230万美元，同比下降12%。 ...

Domestic News - In April, the production of new energy vehicles and charging piles increased by 38.9% and 43.1% respectively [4] - The Buick Envision family is set to roll out its 1.8 millionth vehicle on May 23, achieving a sales growth of 100,000 units in just six months [5] - Equation Leopard announced the delivery of 100,000 vehicles in 18 months, ranking among the top three new force brands for fastest delivery [6] - GAC Haobo launched a mobile service vehicle for trial operation, providing on-site maintenance services for eligible vehicle owners in Beijing and Guangzhou [7] - Chery established a smart technology company with a registered capital of 1.28 billion yuan, enhancing its innovation capabilities in the new energy and smart travel sectors [8] - Changan Automobile Group founded a lightweight technology company with a registered capital of 550 million yuan, focusing on energy management and new energy vehicle sales [10] - Guoxuan High-Tech's G-Yuan solid-state battery has gained recognition from major domestic and international automakers, offering solutions for long-range and high-safety electric vehicles [11] - Guoxin Technology partnered with Aptiv to promote the localization of core automotive chips, enhancing the domestic automotive electronics landscape [12] International News - North America's Q2 automotive production is expected to decrease by approximately 126,000 units due to tariffs [13] - The Ontario provincial government in Canada plans to invest an additional 73 million CAD over the next four years to support the local electric vehicle and battery supply chain [14] - Tata Motors has deployed its first 10 TATA.ev MegaCharger supercharging stations in India, with plans to build up to 500 stations nationwide [15] - Waymo received approval to expand its autonomous taxi services in the San Francisco Bay Area, enhancing its operational footprint [16] Commercial Vehicles - The world's first sodium-ion power battery for commercial vehicles passed rigorous testing, demonstrating significant performance capabilities [18] - Beiqi Foton delivered 300 Truck Mate micro trucks to Nigeria, expanding its market presence in Africa [20] - Feiyue Automobile held a launch event for its new energy product line, showcasing various models designed for green logistics [21] - Suzhou Jinlong's autonomous bus service is set to launch in Yangcheng Shugu, offering free trial rides to the public [22]

Core Viewpoint - The recent surge in Chinese autonomous driving companies expanding internationally highlights their technological advancements and strategic partnerships, particularly with global ride-hailing giant Uber [2][3][4]. Group 1: International Expansion - Chinese autonomous driving companies are increasingly entering international markets, with a focus on L4 autonomous driving as a breakthrough point [3][5]. - WeRide announced a strategic partnership with Uber to expand Robotaxi services to 15 international cities over the next five years, targeting Europe and the Middle East [3][4]. - Momenta is set to introduce autonomous vehicles on Uber's platform in international markets, with commercial operations starting in Europe by early 2026 [4][5]. Group 2: Strategic Collaborations - WeRide is collaborating with various partners, including Swiss authorities and Renault, to launch autonomous driving services in Europe [4][5]. - Momenta's partnership with Uber is seen as a critical step in achieving global-scale autonomous driving, integrating Robotaxi technology into mass-produced vehicles [4][8]. - Xiaoma Zhixing also plans to collaborate with Uber to introduce Robotaxi services in the Middle East, expanding to North America and Southeast Asia [4][5]. Group 3: Market Dynamics - The choice of international markets by Chinese autonomous driving companies reflects a regional differentiation strategy, with Southeast Asia and the Middle East being preferred due to their openness and clear scenarios [5][6]. - The UAE and Saudi Arabia view autonomous driving as a core component of smart city initiatives, driving demand for new energy and unmanned vehicles [6][12]. - The European market is also evolving, with the EU Commission launching an action plan to ensure the automotive industry transitions to smarter and cleaner vehicles [6][12]. Group 4: Technological Advancements - The shift from hardware-centric to software algorithm capabilities is evident as Chinese companies enhance their adaptability to various scenarios [7][9]. - Momenta's "no-map" technology allows for autonomous driving in complex road conditions without relying on high-definition maps, marking a significant advancement in international competitiveness [8][9]. - The development of L2 and L4 technologies is expected to coexist, with Chinese companies achieving high engineering maturity and stability in autonomous driving solutions [9][10]. Group 5: Future Outlook - The penetration rate of L2 and above autonomous driving functions in China is projected to reach 62% by 2025, indicating strong consumer trust in these technologies [10][11]. - The collaboration between domestic suppliers and international partners is expected to enhance the global presence of Chinese autonomous driving technologies [11][12]. - The ongoing development of autonomous driving technology is supported by favorable government policies in various countries, facilitating faster market entry and application [12][13].

来源：环球网 5月20日，第29届世界燃气大会（WGC2025）在北京国家会议中心二期盛大召开。北京北控北斗科技投 资有限公司（北控北斗）与全球领先的无人驾驶企业九识（苏州）智能科技有限公司（九识智能）联合 发布"PPB级燃气管网泄漏高精准无人巡检系统"。该系统首次将PPB级燃气泄漏检测技术与L4级自动驾 驶深度融合，实现燃气管网全天候、全自动、高精度巡检，为全球城市燃气安全管理树立新标杆。 技术突破：PPB级燃气泄漏检测系列装备自主研发 针对城市燃气管网安全巡检业务需求，北控北斗基于中红外激光吸收光谱技术，历时两年时间开展技术 攻关，自主研发PPB级燃气泄漏检测系列装备，推出包括车载式燃气泄漏巡查系统、背包式燃气泄漏核 查设备等重量级产品。该系列设备在巡检过程中能够检测到空气中十亿分之一体积浓度的甲烷和乙烷气 体，实现燃气管道泄漏位置的精准溯源。该系列装备自投入使用以来，已在中国20余个城市累计4万余 公里的燃气管网安全检测工作中发挥重要作用，为城市燃气管网安全运行保驾护航。 创新融合：PPB级燃气泄漏巡检技术首次与L4级无人驾驶技术的深度融合 "让城市燃气安全管理更智能、更高效、更安全"是北控北斗与九识智 ...

每经记者｜刘艳美 每经编辑｜杨欢 近日，浙江省人民政府官网发布《关于支持人工智能创新发展的若干措施》（以下简称《若干措施》），提出到2027年，初步 形成可持续发展、领跑发展格局，培育若干具有全球竞争力和影响力的人工智能企业，全省规模以上人工智能核心产业营业收 入超1万亿元。 《若干措施》指出，打造高能级万亿产业生态，加快构建以杭州市、宁波市为核心，其他地区特色化差异化发展，覆盖芯片、 服务器、系统软件、云平台、大模型、智能终端等全产业链的人工智能发展格局。布局省级人工智能产业平台（园区），大力 引育处于产业链关键环节、拥有自主核心技术的高成长性企业。鼓励有条件的地方对企业研发费用按比例给予财政补助。研究 大模型前沿问题的保护策略，推动从"技术主导"向"规则主导"转变。 解读：当下，全球人工智能技术加速迭代，成为推动科技跨越式发展、产业优化升级和生产力整体跃升的驱动引擎。而今年以 来，"杭州六小龙"爆火出圈，让浙江人工智能实力备受瞩目。 去年11月发布的《2024年浙江省人工智能产业发展报告》（以下简称《发展报告》）显示，浙江人工智能综合发展水平居北 京、广东、上海、江苏之后，位列全国第五，人工智能产业年产值 ...

Core Points - The visit of a Japanese youth delegation to Beijing highlights the importance of cultural exchange and friendship between Japan and China [2][9] - The delegation experienced modern China's rapid development and technological advancements, particularly in logistics and autonomous driving [3][4][6] Group 1: Cultural Exchange - The Japanese youth delegation, led by Inoue Masayoshi, visited Beijing from May 12 to 16, 2023, to engage with Chinese youth and experience the cultural blend of the city [2] - Participants expressed a deep emotional connection to China, with many sharing their positive experiences and the warmth of Chinese hospitality [3][4] - The visit aimed to foster understanding and friendship between Japanese and Chinese youth, with hopes for future exchanges [9] Group 2: Technological Advancements - The delegation visited JD Logistics' Asia No.1 smart industrial park, which features over 100 AGV robots for efficient logistics operations [3] - The park's drone technology, capable of carrying 10 to 15 kilograms, was of particular interest to the students, showcasing innovative logistics solutions [3] - The Beijing Economic-Technological Development Area has established a high-level autonomous driving demonstration zone, attracting over 30 testing companies and accumulating over 35 million kilometers of testing mileage [6][7]

Group 1: Hong Kong Stock Market Highlights - Midea Group opened up 2% against the market trend after being included in the Hang Seng Index [1] - Yuan Da Pharmaceutical opened over 3% higher, with its first "zero radiation" nuclear medicine factory set to commence production in June [1] - Bruker surged over 6% after being added to the Hang Seng Composite Index, with institutions stating it meets the criteria for inclusion in the Hong Kong Stock Connect [1] - Xuanwu Cloud rose over 22% following a strategic partnership with Tencent Cloud, focusing on AI-driven retail and cloud communication growth [1] - Mixue gained over 5% after being included in the Hang Seng Composite Index, with expectations of being added to the Hong Kong Stock Connect next month [1] - Konnate Optical increased nearly 3% as it plans to invest $4 million in establishing a resin lens production line in Japan [1] - Solar stocks declined, with Q1 solar product exports dropping over 30%, indicating potential pressure on the industry chain [1] - Weilang Delicious rose over 5% after completing a placement that raised over HKD 1.1 billion, with CICC noting the scale of the placement is relatively controllable [1] - Logistics stocks continued to rise, with notable increases in shares of Guangdong-Hong Kong-Macau Holdings, Guangdong Transportation, Kerry Logistics, and ZTO Express [1] Group 2: US Stock Market Highlights - Pony.ai surged nearly 14%, with expectations of achieving positive gross margins for complete vehicles this year [2] - Walmart increased nearly 2% as the company plans to raise prices on certain products in the US [2] - QUBT rose over 39% after turning profitable in Q1 [2] - CoreWeave jumped over 22%, with NVIDIA holding a 7% stake [2] - Bilibili increased over 2%, with Goldman Sachs projecting a 71% year-on-year growth in gaming revenue for Q1 [2] - Hesai Technologies rose over 6% at one point, closing nearly 2% higher amid rumors of a secret application for a Hong Kong listing [2] - Novo Nordisk fell nearly 3% following the announcement of the CEO's departure [2] - Applied Materials (AMAT) dropped over 5% as Q2 revenue fell short of expectations, with a bleak outlook for Q3 [2]

据广州日报报道，记者从部分银行及房产中介获悉，5月17日起，广州地区银行首套房贷利率从原来的LPR-60个基点调整为LPR-50个 基点，即五年期以上首套房贷利率上涨10BP（基点），由原来的3.0%上调到3.1%。 点评：广州首套房贷利率微幅上调，反映出在LPR（贷款市场报价利率）下调背景下银行对利差的调整。这将轻微增加购房成本，或 对刚需及市场预期形成一定抑制，需关注后续市场反应。 ｜2025年5月19日星期一｜ NO.1 广州五年期以上首套房贷利率上调10BP 点评：旅客会在购物、餐饮、住宿、娱乐等方面产生大量消费，从而刺激香港本地消费市场，带动相关产业发展，为香港经济增长注 入动力。 NO.4 广州将开通8条自动驾驶专线 据"中国广东发布"微信公众号5月17日消息，广州将在市中心开通8条自动驾驶出行服务示范运营专线，开启全国首个全天时、全天候 的自动驾驶出行服务。据悉，服务路线覆盖广州塔、广交会展馆、花园酒店、珠江新城等地标，以及广州白云国际机场、广州南站等 交通枢纽，覆盖了广州人流、车流、物流最密集繁忙的区域。 点评：作为全国首个开启全天时、全天候自动驾驶出行服务的城市，广州展现出了在科技创新和城市 ...

Core Viewpoint - The article discusses the advancements in the field of autonomous driving, particularly focusing on the Diffusion model and its application in generating driving trajectories, highlighting the differences between VLM and VLA systems [1][4]. Group 1: Diffusion Model Explanation - Diffusion is a generative model that learns data distribution through a process of adding noise (Forward Process) and removing noise (Reverse Process), akin to a reverse puzzle [4]. - The model's denoising process involves training a neural network to predict and remove noise, ultimately generating target data [4]. - Diffusion not only generates the vehicle's trajectory but also predicts the trajectories of other vehicles and pedestrians, enhancing decision-making in complex traffic environments [5]. Group 2: VLM and VLA Systems - VLM consists of two systems: System 1 mimics learning to output trajectories without semantic understanding, while System 2 has semantic understanding but only provides suggestions [2]. - VLA is a single system with both fast and slow thinking capabilities, inherently possessing semantic reasoning [2]. - The output of VLA is action tokens that encode the vehicle's driving behavior and surrounding environment, which are then decoded into driving trajectories using the Diffusion model [4][5]. Group 3: TransDiffuser Architecture - TransDiffuser is an end-to-end trajectory generation model that integrates multi-modal perception information to produce high-quality, diverse trajectories [6][7]. - The architecture includes a Scene Encoder for processing multi-modal data and a Denoising Decoder that utilizes the DDPM framework for trajectory generation [7][9]. - The model employs a multi-head cross-attention mechanism to fuse scene and motion features during the denoising process [9]. Group 4: Performance and Innovations - The model achieves a Predictive Driver Model Score (PDMS) of 94.85, outperforming existing methods [11]. - Key innovations include anchor-free trajectory generation and a multi-modal representation decorrelation optimization mechanism to enhance trajectory diversity and reduce redundancy [11][12]. Group 5: Limitations and Future Directions - The authors note challenges in fine-tuning the model, particularly the perception encoder [13]. - Future directions involve integrating reinforcement learning and referencing models like OpenVLA for further advancements [13].