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 理想汽在ICCV'25期间也分享了些新东西！目前还没有视频对外。 VLA团队负责人詹锟老师做了一场世界模型的presentation，名为World Model: Evolving from Data Closed-loop to Training Closed-loop。自动驾驶之心第一时间做了解 读分享给大家~ 首先是介绍下理想VLA司机大模型： 回顾了理想汽车智能驾驶的发展路线，从规则时代的轻图和无图，再到基于AI的E2E+VLM快慢双系统和VLA， 这四个方案中Nav（导航）是重点突出的模块。 下面介绍的是数据闭环的价值。左上角这张图是一个完整的数据闭环流程： 影子模式验证→经由数据触发回传到云端进行数据挖掘→有效样本进行自动标注→生 成训练集训练模型→模型下发验证性能。 这个过程已经可以做到一分钟的数据回传。 目前已经有15亿公里的驾驶数据，200+的Trigger来生产15-45s的Clip数据。 目前理想的端到端量产版本MPI已经到了220+， ...

完成双线布局后，理想重回"创业模式" 从理想ONE到理想L系列，增程路线为理想汽车打开了家庭用车市场的大门；而随着理想MEGA与理想i 系列构建起完整的纯电矩阵，并且MEGA在高端市场取得突破，理想i8和i6累计订单也突破10万辆，理 想汽车终于实现了"两条腿走路"。当增程和纯电双能源形式的产品布局正式成型，理想汽车的"上一个 十年"完美收官，是时候全身心投入下一个战场了。 于是，理想汽车做出了企业愿景的升级——要成为"全球领先的具身智能企业"。企业愿景的转变意味着 产品定位的根本性变革，从此汽车不再是单纯的交通工具，也不是简单的智能终端，而是物理世界中的 具身智能机器人。李想对此有着深刻的思考："电动车的选择不是不好，而是不够；智能终端的选择也 不是不好，而是不够；只有选择具身智能这个最难的题，我们才可以真正改变用户的生活。" 愿景升级，必然倒逼组织进化。从2025年第四季度起，理想果断回归"创始人模式"：李想亲自掌管产 品、公司战略、人才管理，将决策链条大幅缩短。过去三年，理想曾努力向职业经理人模式靠拢，但实 践证明，"我们变成了越来越差的自己"李想说。在AI技术高速迭代、行业格局未定的当下，这种强调流 ...

Core Insights - Li Auto reported a revenue of 27.4 billion yuan for Q3 2025, with a total revenue of 83.5 billion yuan for the first three quarters, leading the "new force" car manufacturers [1] - The company has successfully expanded its pure electric vehicle lineup with the launch of the Li i8 and Li i6, achieving over 100,000 total orders [2] - Li Auto is focusing on AI and embodied intelligence technology to drive future growth, with a projected R&D expenditure of 12 billion yuan in 2025 [4][5] Financial Performance - In Q3 2025, Li Auto achieved a revenue of 27.4 billion yuan, contributing to a total of 83.5 billion yuan for the first three quarters [1] - The company has a cash reserve of 98.9 billion yuan as of the end of September, providing a strong financial foundation for ongoing investments [4] Product Development - The launch of the Li i8 and Li i6 has allowed Li Auto to cover mainstream and high-end family electric vehicle markets, establishing a solid foundation for long-term growth [3] - The Li i8 is the company's first pure electric six-seat SUV, while the Li i6 targets younger consumers as a new five-seat electric SUV [2] Market Strategy - Li Auto is expanding its self-built 5C supercharging network, which has surpassed 3,500 stations as of October, to enhance charging services for its electric products [2] - The company is adopting a dual-supplier model for the Li i6 battery supply to address production capacity challenges, with expectations to increase monthly production to 20,000 units by early next year [2] AI and Technology Investment - Li Auto is investing heavily in AI and embodied intelligence, with over 6 billion yuan allocated to AI research in 2025 [4][5] - The company has opened its self-developed intelligent vehicle operating system, "Star Ring OS," with over 55 potential partners and 16 signed agreements to foster innovation in the industry [4] Future Outlook - Li Auto's CEO expressed confidence in transforming user experiences through embodied intelligence, positioning the company for significant revenue growth in the coming years [5] - Industry experts believe that Li Auto's focus on AI technology and its established brand recognition will facilitate its transition from a range-extended vehicle leader to a strong player in the pure electric market [3][5]

Core Insights - Li Auto reported Q3 revenue of 27.4 billion yuan, leading the new energy vehicle sector, with total revenue for the first three quarters reaching 83.5 billion yuan [1] - As of the end of Q3, Li Auto's cash reserves stood at 98.9 billion yuan, indicating strong financial health [1] - The company successfully launched two new electric SUV models, Li Auto i8 and i6, with total orders exceeding 100,000 units [1] - Li Auto's R&D expenses for Q3 reached 3 billion yuan, with an expected annual investment of 12 billion yuan, including over 6 billion yuan in artificial intelligence [1] - The VLA driver model, utilizing world models and reinforcement learning, achieved a monthly usage rate of 91% in October, showcasing industry-leading user penetration [1] Product and Technology Strategy - Li Auto is leveraging both range-extended and pure electric strategies, alongside self-developed chips and global expansion, to enhance its technological capabilities [2] - The company aims to lead the market through product and technology innovation, focusing on quality to improve user experience and drive sustainable growth [2] Financial Performance and Management Vision - Excluding the estimated recall costs for the 2024 Li Auto MEGA, the gross margin for Q3 was 20.4%, reflecting the company's operational resilience [3] - Li Auto's CEO expressed a commitment to reorganizing at the organizational, product, and technological levels to ensure the company can navigate cycles and lead in technology, creating unique value for users and society [3]

Group 1 - The 23rd Guangzhou International Auto Show, themed "New Technology, New Life," commenced on November 21, showcasing 1,085 vehicles, with 93 new car debuts [1] - Among the exhibited vehicles, 629 were new energy vehicles (NEVs), accounting for nearly 60% of the total, reflecting a significant increase from the previous year's show where NEVs did not exceed half [1] - In 2024, China's automotive production and sales are expected to exceed 30 million units, with NEV production and sales surpassing 10 million units for the first time, and NEV ownership increasing 5.4 times compared to 2020 [1] Group 2 - The auto show highlighted the rapid development of electric and intelligent vehicles, with a focus on advanced driving technologies becoming a core attraction [2] - Companies like Avita and Huawei announced a new phase of collaboration, integrating advanced driving solutions and intelligent cockpit technologies into Avita models [2] - The exhibition saw an 81% increase in the number of exhibitors showcasing innovations in electric motors, energy storage, autonomous driving, and artificial intelligence, indicating a collaborative advancement across the entire automotive supply chain in China [2][3]

Core Insights - The article discusses the advancements in autonomous driving technology, particularly focusing on the transition from data closed-loop systems to training closed-loop systems, marking a new phase in autonomous driving development [18][21]. Group 1: Development of Ideal Auto's Intelligent Driving - Ideal Auto's intelligent driving has evolved through various stages, from rule-based systems to AI-driven E2E+VLM dual systems and VLA, with a strong emphasis on navigation as a key module [6]. - The current end-to-end mass production version of MPI has reached over 220, representing a 19-fold increase compared to the version from July 2024 [13]. Group 2: Data Closed-Loop Value - The data closed-loop process includes shadow mode validation, data feedback to the cloud for mining, automatic labeling of effective samples, and model training, with data return achievable in one minute [9][10]. - Ideal Auto has accumulated 1.5 billion kilometers of driving data, utilizing over 200 triggers to produce 15-45 second clip data [11]. Group 3: Transition to Training Closed-Loop - The core of the L4 training loop involves VLA, reinforcement learning (RL), and world models (WM), optimizing trajectories through diffusion and reinforcement learning [23]. - Key technologies for closed-loop autonomous driving training include regional simulation, synthetic data, and reinforcement learning [24]. Group 4: Reconstruction and Generation Work - Ideal Auto has made significant progress in reconstruction and generation, with multiple top conference papers published in the last two years [28][32][34]. - The generation applications range from scene editing to scene migration and scene generation [36]. Group 5: Interactive Agents and System Capabilities - The development of interactive agents is highlighted as a critical challenge in the training closed-loop [40]. - System capabilities are enhanced through world models providing simulation environments, diverse scene construction, and accurate feedback from reward models [41]. Group 6: Community and Collaboration - The article mentions the establishment of nearly a hundred technical communication groups related to various autonomous driving technologies, with a community of around 4,000 members and over 300 companies and research institutions involved [50][51].

Core Insights - The article discusses the advancements in autonomous driving technology, particularly focusing on the transition from data closed-loop systems to training closed-loop systems, marking a new phase in autonomous driving development [17][20]. Group 1: Development of Li Auto's VLA Model - Li Auto's VLA driver model has evolved through various stages, from rule-based systems to AI-driven E2E+VLM systems, with a strong emphasis on navigation as a key module [6]. - The end-to-end mass production version of MPI has reached over 220 units, representing a 19-fold increase compared to the version from July 2024 [12]. Group 2: Data Closed-Loop Value - The data closed-loop process includes shadow mode validation, data mining in the cloud, automatic labeling of effective samples, and model training, with a data return time of one minute [9][10]. - Li Auto has accumulated 1.5 billion kilometers of driving data, utilizing over 200 triggers to produce 15-45 second clip data [10]. Group 3: Transition to Training Closed-Loop - The core of the L4 training loop involves VLA, reinforcement learning (RL), and world models (WM), optimizing trajectories through diffusion and reinforcement learning [22]. - Key technologies for closed-loop autonomous driving training include regional simulation, synthetic data, and reinforcement learning [24]. Group 4: Simulation and Generation Techniques - Simulation relies on scene reconstruction, including visual and Lidar reconstruction, while synthetic data generation utilizes multimodal techniques [25]. - Li Auto's recent advancements in reconstruction and generation have led to significant improvements, with multiple top conference papers published in the last two years [26][29][31]. Group 5: Interactive Agents and System Capabilities - The development of interactive agents is highlighted as a critical challenge in the training closed-loop [37]. - System capabilities are enhanced through world models providing simulation environments, diverse scene construction, and accurate feedback from reward models [38]. Group 6: Community and Collaboration - The article mentions the establishment of nearly a hundred technical discussion groups related to various autonomous driving technologies, with a community of around 4,000 members and over 300 companies and research institutions involved [44][45].

Core Viewpoint - Li Auto's CEO Li Xiang emphasizes the necessity of advanced driver assistance systems (ADAS) without usage fees, positioning the upcoming Li Auto i6 as a leader in smart electric vehicles [1] Group 1: Company Strategy - Li Auto will launch the new five-seat pure electric SUV, Li i6, on September 26, featuring the AD Max system as standard and promising no usage fees [1] - The company aims to enhance consumer access to advanced AI technology from China by eliminating fees associated with high-level driving assistance features [1] Group 2: Industry Context - The competition in the new energy vehicle sector has shifted towards smart technology, with advanced driver assistance being a key value point [1] - Li Xiang critiques the current industry practice of charging for advanced driving features, which he believes limits user experience [1] Group 3: Product Features - The Li i6 is built on an 800V high-voltage platform, allowing for a rapid charge of 10 minutes to achieve a range of 500 kilometers, with a maximum CLTC range of 720 kilometers [1] - Li Xiang claims that the AD Max system's performance is among the best in the industry, with most consumers recognizing it as one of the top two systems available [1]

Core Viewpoint - The founder and CEO of Li Auto, Li Xiang, has publicly refuted claims circulating online regarding the "bloodied version" of the OTA 8.0 update for the Li VLA driver model, urging the public not to believe or spread misinformation [1] Group 1 - Li Auto's CEO addressed misinformation regarding the OTA 8.0 version of the Li VLA driver model [1] - The company emphasizes the importance of verifying information before sharing it [1]

Group 1 - The CEO of Li Auto, Li Xiang, addressed misinformation circulating online regarding the recently launched OTA 8.0 version of the Li Auto VLA driver model, urging the public not to believe or spread these claims [1] - Li Xiang expressed gratitude to the community for their support amidst the confusion surrounding the "bloodied version" allegations [1] Group 2 - The company is actively managing its public relations to counteract negative narratives about its product updates [1] - The emphasis on clarifying misinformation indicates the company's commitment to maintaining its reputation and customer trust [1]