2025年11月25日（周二）报纸版的人民日报共有5版有广告。 | 01版:变闻 | 02版:要闻 | 03版:变闻 | 04版:要闻 | | --- | --- | --- | --- | | 05版:评论 | 06版:蛋闻 | 07版:文化 | 08版:广告 | | 09版:理论 | 10版:综合 | 11版: 固际 | 12版:广告 | | 13版:特别报道 | 14版:特别报道 | 15版:特别报道 | 16版:广告 | | 17版:国际 | 18版:新农村 | 19版:广告 | 20版:广告 | 作为对比2025年11月24日（周一）有1版有广告。 | 01版:变闻 | 02版:要闻 | 03版:变网 | 04版:要闻 | | --- | --- | --- | --- | | 05版:评论 | 06版:要闻 | 07版:读者来信 | 08版:广告 | | 09版:理论 | 10版:社会 | 11版:生态 | 12版:国际 | | 13版:特别报道 | 14版:特别报道 | 15版:特别报道 | 16版:新青年 | | 17版:国际 | 18版:财经 | 19版:绿色 | 20版:副刊 | | 01 ...

Core Insights - The Chinese automotive industry has achieved significant advancements over the past decade, particularly in the field of new energy vehicles (NEVs), which have become a vital component for high-quality development [1] - Li Auto, established in 2015, has emerged as a leading mid-to-high-end NEV brand in China, achieving annual sales of 500,000 vehicles and over 100 billion yuan in revenue for two consecutive years [1][2] - The company is committed to innovation and has developed a robust supply chain ecosystem, known as "Li Chain," which integrates nearly a thousand partners to enhance collaborative growth [4][5] Industry Development - The NEV sector in China is projected to exceed 12 million units in annual production and sales by 2024, with Li Auto reaching its milestone of 1 million vehicles produced in just 58 months [2] - Li Auto is actively contributing to the development of Changzhou as "China's New Energy Capital," aiming for the local NEV industry to surpass 850 billion yuan in scale by 2024 [2] Technological Innovation - During the 14th Five-Year Plan period, Li Auto has focused on becoming a global leader in artificial intelligence (AI) and plans to invest over 6 billion yuan in AI development this year [3] - The company has launched the VLA driver model and "Li Xiang Classmate" AI system, marking its entry into a new phase of AI-driven development [3] Supply Chain and Ecosystem - Li Auto has established a supply chain system characterized by "excellent growth, intelligent innovation, and green health," with procurement amounts growing from billions to trillions in just three years [4][5] - The "Li Chain" ecosystem promotes localized industrial collaboration, with 80% of suppliers located in the Yangtze River Delta region, and 50% concentrated in Jiangsu [5] Talent Development - Li Auto has partnered with over a hundred universities to cultivate more than 5,000 high-end industry talents through initiatives like the "Li Xiang+" talent program [6] Future Outlook - The company aims to leverage technological innovation to drive high-quality industry development and enhance user experiences in green and smart mobility [8]

Core Viewpoint - The article discusses the shift in the autonomous driving industry from a data-driven approach to a training-driven approach, emphasizing the importance of world models and reinforcement learning in achieving Level 4 (L4) autonomy [2][4][6]. Group 1: Transition from Data Loop to Training Loop - The current data loop is insufficient for advancing autonomous driving technology, necessitating a shift to a training loop that allows for continuous model iteration through environmental feedback [4][11]. - Ideal's approach involves building a world model training environment in the cloud, which integrates prior knowledge and driving capabilities into the vehicle's VLA model [11][30]. - The world model encompasses environment construction, agent modeling, feedback mechanisms, and various scenario simulations, which are crucial for the training loop [13][31]. Group 2: Simulation and Evaluation Techniques - Ideal employs a combination of reconstruction and generation techniques for simulation, allowing for both stable and dynamic outputs [14][15][16]. - The Hierarchy UGP model, developed in collaboration with academic institutions, achieves state-of-the-art results in large-scale dynamic scene reconstruction [21][19]. - The focus on synthetic data generation enhances the diversity and complexity of training scenarios, improving model performance [25][24]. Group 3: Reinforcement Learning and Challenges - The reinforcement learning world engine enables models to explore training environments and receive feedback, with five key factors influencing its effectiveness [25][27]. - The simulation of interactions between multiple agents poses significant challenges, with Ideal exploring self-play and reward function adjustments to enhance sample diversity [27][29]. Group 4: Commercialization and Technological Advancements - Ideal has successfully established a profitable business model, which supports its ongoing research and development efforts, with over 10 billion yuan invested in the self-developed Star Ring OS [32][33]. - The Star Ring OS enhances vehicle performance by streamlining communication between different control systems, significantly reducing braking distances [35][36]. - The open-source initiative of the Star Ring OS is expected to benefit the entire industry, reducing development costs for other automakers [39][40]. Group 5: Industry Position and Future Outlook - Ideal is positioning itself as a leading player in the AI-driven automotive sector, with a focus on becoming a "space robotics company" [48][50]. - The company has established a research-production closed loop, allowing for rapid application of research findings to production, exemplified by the DriveVLM project [52]. - The article concludes that while many companies are investing in AI and robotics, few have achieved the comprehensive capabilities demonstrated by Ideal and Tesla [53].

Core Viewpoint - Zhao Dong is recognized as a key figure in ensuring information security for new energy smart connected vehicles, emphasizing the importance of a robust automotive operating system for the industry's development [3][4]. Group 1: Background and Experience - Zhao Dong has extensive experience in information security, having worked in the field for decades and holding a master's degree from Tsinghua University in electrical engineering [4]. - He made a strategic decision to enter the new energy smart connected vehicle sector in 2014, believing it to be a crucial area for national development despite its nascent stage at the time [4]. Group 2: Security Framework Development - Under Zhao Dong's leadership, the team established a comprehensive defense system for automotive information security, integrating hardware and software to enhance security capabilities [5][6]. - The security philosophy is embedded in the Integrated Product Development (IPD) process, ensuring safety measures are included from the design phase through to testing [6]. Group 3: AI Integration and Innovation - Zhao Dong has proactively integrated AI technology into core security processes, significantly improving operational efficiency in vulnerability management and intrusion detection [3][8]. - The intelligent security platform developed under his guidance has become central to the company's defense strategy, utilizing big data and AI for proactive threat detection and response [8]. Group 4: Industry Contributions and Standards - Zhao Dong actively participates in industry standards development and has shared his expertise at various national and industry forums, earning high praise for his contributions [11][13]. - The open-source release of the self-developed operating system, Xinghuan OS, marks a significant milestone for the industry, allowing for community collaboration to enhance security features [10][12]. Group 5: Talent Development - Zhao Dong emphasizes talent cultivation within the organization, regularly organizing training and knowledge-sharing sessions to enhance the skills of team members [13].

Core Viewpoint - The article discusses how Li Auto is transforming its cockpit experience through the development of its self-developed AI model "Mind GPT," positioning itself as a leader in the intelligent cockpit space amidst increasing competition in the automotive industry [4][5][6]. Group 1: Development of AI Capabilities - Li Auto has shifted from relying on third-party suppliers for its voice assistant to developing its own AI capabilities, marking a significant transformation in its cockpit technology [6][8]. - The company aims to enhance user interaction through the "Li Xiang Classmate" app, which integrates the Mind GPT model into its vehicle systems, allowing for more natural and efficient user engagement [4][23]. - The internal team was formed to regain data ownership and establish a self-sufficient AI development environment, which has led to significant improvements in user experience and interaction [14][15]. Group 2: Strategic Vision and Implementation - Li Auto's CEO, Li Xiang, emphasizes the necessity of developing large models to compete effectively in the AI space, stating that without them, the company cannot be considered an AI company [5][19]. - The company has set ambitious goals, including becoming a leading AI enterprise by 2030, which reflects a shift in its identity from a traditional car manufacturer to a technology-driven company [19][28]. - The introduction of the Mind GPT model is part of a broader strategy to integrate AI into various aspects of the user experience, including travel assistance, entertainment, and education [23][24]. Group 3: Future Directions and Innovations - Li Auto is focusing on the development of a foundational model that will support its AI initiatives, with plans to open-source its operating system to enhance collaboration and reduce costs [31][32]. - The company envisions vehicles evolving from mere transportation tools to AI-driven "space robots," indicating a significant shift in the automotive landscape [32][33]. - The establishment of an AI committee aims to oversee technological advancements and investment decisions, ensuring that the company remains at the forefront of AI innovation in the automotive sector [27][30].

Core Viewpoint - The article discusses the transition of automotive electronics from a "multi-ECU distributed" architecture to a "centralized" one, highlighting the benefits of reduced hardware and concentrated resources, while also addressing the risks of potential interference between functions of different safety levels on the same computing platform [8][9]. Group 1: Background and Objectives - The shift to centralized architecture increases the risk of "safety crosstalk," where one function may inadvertently alter another's data, potentially leading to malfunctions [9]. - The goal of the intelligent vehicle control OS is to establish clear spatial and permission boundaries for integrated functions, ensuring stable coexistence of different safety levels on the same platform [10]. Group 2: Key Features of the Isolation Framework - The intelligent vehicle control OS introduces a lightweight safety isolation framework that emphasizes three main features: hard isolation, low overhead, and fast recovery [10]. - Hard isolation involves a multi-dimensional memory mapping and fine-grained isolation mechanism that utilizes hardware Memory Protection Units (MPU) to protect application tasks and data [12][25]. - Low overhead is achieved through a lightweight synchronous remote call mechanism that decouples memory access domain switching from task scheduling, allowing for efficient inter-application communication with minimal latency [15][18]. - Fast recovery is facilitated by a fault detection and recovery mechanism that allows for independent reset of isolated units without affecting other applications, thus maintaining system stability [19][30]. Group 3: Technical Solutions - The lightweight software decoupling framework supports spatial isolation mechanisms across cores, system software, and application layers, balancing safety and resource efficiency [22][24]. - The multi-dimensional layered memory mapping allows for precise data allocation and classification based on ownership, functionality, and software hierarchy [25][27]. - The high-performance communication mechanism ensures that calls between isolated functional units maintain task context and minimize resource consumption [28][30]. Group 4: Practical Implementation - The article mentions practical demonstrations using TC397 or E3650 development boards, showcasing the collaborative effects of hard isolation, low overhead communication, and fast recovery in real deployment scenarios [37]. - The recovery process involves a series of steps from fault detection to resource cleanup and application restart, ensuring that unaffected applications continue to operate normally [38]. Group 5: Conclusion - The intelligent vehicle control OS effectively addresses the challenges of crosstalk and real-time performance in centralized vehicle control by implementing a lightweight safety isolation framework, achieving a balance between safety and efficiency [40].

Core Viewpoint - Li Auto demonstrates resilience and strength in the face of market competition and internal adjustments, achieving robust financial performance in Q2 2023 [1] Financial Performance: Steady Profitability and Cash Reserves - The company reported a quarterly revenue of 30.2 billion yuan, a 16.7% increase from Q1 [1] - Net profit for the quarter reached 1.1 billion yuan, a 69.6% increase from Q1, marking 11 consecutive quarters of profitability [2] - Operating profit grew by 204.4% from Q1, reaching 827 million yuan [2] - As of Q1 2025, cash reserves stood at 110.7 billion yuan, supporting long-term R&D investments [2] R&D Investment: Significant Funding and Technological Leadership - Li Auto maintains a high R&D investment strategy, with quarterly R&D expenses of 2.8 billion yuan, averaging 1 billion yuan every three days [3] - Total expected R&D expenditure for the year is 12 billion yuan, with over 6 billion yuan allocated to AI technology [3] - The company has made significant breakthroughs in smart and electric technologies, with the upcoming OTA 8.0 featuring AI-based advanced driving and smart cockpit technologies [3] - The self-developed VLA driver model enhances user experience through natural language interaction and memory of user preferences [3] Product Strategy: Expansion into Pure Electric Vehicles - Li Auto is actively expanding its product line from range-extended electric vehicles to pure electric models, with the Li i8 SUV launched in July 2025 [5] - The Li i8 features long range, low energy consumption, spacious design, and low drag coefficient, having undergone extensive testing [5] - The company plans to release the Li i6 in September, targeting the mainstream market priced between 200,000 to 350,000 yuan [5] Charging Infrastructure: Accelerated Development - To support the launch of pure electric models, Li Auto is rapidly expanding its charging network, with 3,050 supercharging stations and 16,795 supercharging piles established by August 10, 2025 [6] - Both the Li i8 and Li i6 can achieve a charging experience of 500 kilometers in just 10 minutes, ensuring a worry-free electric driving experience [6]

Group 1 - Xiaomi's YU7 achieved over 200,000 pre-orders in just three minutes, setting a new benchmark in the smart car industry, which is seen as a "myth" that few can replicate [1] - The success of YU7 is attributed to its emotional value, appealing design, competitive pricing, and strong marketing by Lei Jun, which resonated with consumers [1][2] - Other companies, such as XPeng, are also focusing on emotional value in their products, with XPeng's new P7 emphasizing both technology and aesthetics [2][3] Group 2 - The concept of emotional value is becoming central to the strategies of various automakers, with XPeng's founder emphasizing the importance of aesthetics alongside technology [2][3] - Companies like Li Auto, Leap Motor, and Great Wall are also working on enhancing emotional value in their products, indicating a broader industry trend [3][19] - The shift towards emotional value reflects a change in consumer preferences from basic transportation needs to a desire for personalized and emotionally resonant products [20] Group 3 - Cross-domain integration is identified as a key factor in enhancing emotional value in smart vehicles, requiring collaboration across various functional domains [3][21] - XPeng and Volkswagen's collaboration on the CEA architecture aims to support both electric and hybrid vehicles, showcasing the industry's move towards integrated systems [24] - Companies are investing in advanced computing architectures to facilitate cross-domain integration, which is essential for delivering emotional value in smart cars [25][28]

Core Viewpoint - The article discusses the development and vision of the Li Auto Star Ring OS, emphasizing its role in the evolution from smart vehicles to space robots, and the importance of open-source collaboration in achieving technological advancements in the automotive industry [1][3]. Group 1: Background of Li Auto Star Ring OS - The Star Ring OS was announced as an open-source project on March 27, 2024, and has since undergone two significant version releases, gaining industry recognition [1]. - Smart vehicles are seen as the initial successful commercial application of space robots, serving as a critical step towards achieving general-purpose space robots [5][6]. Group 2: Trends in Space Robot Technology Development - The ultimate goal is to develop general-purpose space robots, with the feasibility proven by the success of smart vehicles and breakthroughs in core technologies like VLA (Vision-Language-Action Models) [7][9]. - Key trends include the pursuit of extreme hardware-software collaboration, centralized and efficient utilization of computing resources, system security as a foundational prerequisite, and embracing open-source for efficient co-construction [11][12]. Group 3: Issues with Classic Automotive Software Solutions - The classic automotive software architecture, which emerged in 2000, has led to an explosion in the number of electronic control units (ECUs), resulting in integration difficulties and poor reusability [15][18]. - The limitations of the classic approach include a modular design that creates information silos and a collaborative model that is open yet not fully open-source, hindering innovation and efficiency [18][20]. Group 4: Open Source Solutions - The Star Ring OS is designed as a cohesive system with four core components: AI computing system, intelligent real-time system, communication middleware, and information security system [25][26]. - The OS aims to provide a unified, collaborative, flexible, and secure digital foundation for space robots, addressing the identified contradictions in the development process [29][30]. Group 5: Open Source Ecosystem Construction - The open-source ecosystem aims to build a unified, open, and general intelligent system foundation for the space robot era, promoting industry collaboration and reducing costs [36][39]. - Achievements include significant cost savings, reduced development cycles, and enhanced performance, with the platform enabling efficient coordination and flexible deployment of vehicle control systems [40][41]. Group 6: Future Work Focus - Future efforts will concentrate on community building, expanding ecosystem partnerships, and enhancing industry influence, with a clear path towards open governance and robust technology development [44][48]. - The next steps include supporting new chip platforms, enhancing core capabilities, and providing efficient community development facilities [49][50].

Core Viewpoint - The article emphasizes the advancements in Li Auto's VLA (Vehicle Language Assistant) technology, highlighting its capabilities in intelligent driving and AI security, as well as the company's leadership role in the newly established Automotive AI Standardization Promotion Center. Group 1: VLA Technology Advancements - Li Auto's VLA can understand and execute voice commands for driving tasks, enhancing user experience by integrating visual, linguistic, and behavioral capabilities into a single chip [4][6]. - The VLA system employs a newly designed architecture that combines end-to-end models with logical reasoning, optimizing driving decisions and trajectory predictions to improve vehicle performance in complex environments [6][7]. - The VLA incorporates user interaction features, allowing drivers to communicate their needs easily, such as finding locations or adjusting driving speed [7][9]. Group 2: AI Security Measures - Li Auto has established a comprehensive security framework that covers vehicles, cloud services, apps, and charging networks, creating a robust defense system against potential threats [9][12]. - The article discusses the complexity of AI-targeted attacks, which can manipulate AI decision-making through subtle inputs, necessitating advanced defensive strategies [9][11]. - Li Auto is focusing on building an AI security capability system that addresses adversarial attacks, safety alignment, and behavioral constraints through continuous innovation and validation [11][12]. Group 3: Standardization Initiatives - The Automotive AI Standardization Promotion Center has been established to focus on AI safety management, risk governance, and the development of international standards for automotive AI [13][14]. - Li Auto has taken a leadership role in two research groups within the center, aiming to integrate research outcomes into products to enhance safety and intelligence in driving experiences [16][12]. - The SAFER AI initiative has been launched to create a comprehensive framework for AI safety assessment, engineering research, and standard development in the automotive sector [14][16].