Core Insights - The 2025 FISITA Intelligent Safety Conference aims to promote global collaboration in autonomous driving safety technology, focusing on key themes such as functional safety, testing evaluation, connected cooperative driving, information security, and artificial intelligence [1] Group 1: Functional Safety and Risk Assessment - The introduction of the SOTIF (Safety of the Intended Functionality) Index 1.0 aims to address the rising number of autonomous driving accidents by establishing a comprehensive risk and capability evaluation system [3] - The SOTIF Index is designed to create a multidimensional, quantifiable measurement system to assess the maturity and confidence level of autonomous driving systems in handling expected functional safety [3] - The need for a scientific and quantifiable evaluation paradigm is emphasized to ultimately reduce the accident rate to zero [3] Group 2: Challenges in Autonomous Driving - Various factors such as different road and weather conditions increase the accident rate of autonomous vehicles, posing higher reliability challenges for Advanced Driver Assistance Systems (ADAS) [4] - The necessity to establish mechanisms for identifying unknown risks and quantifying uncertainties in AI systems is highlighted [4] - Strict adherence to traffic regulations is essential, as violations contribute significantly to accident rates [4] Group 3: Industry Development and Collaboration - The penetration rate of L2-level assisted driving in passenger vehicles reached 62.1% in the first half of the year, with steady progress in L3/L4 autonomous driving testing [5] - The importance of top-level design and policy development is stressed to address new safety challenges and enhance global and industrial collaboration [5] - The establishment of a safety assurance system is crucial for promoting healthy industry development [5] Group 4: Integrated Safety Mechanisms - The need for an integrated safety approach that combines functional safety and cybersecurity is emphasized, moving beyond traditional protective mechanisms [8] - The concept of a "traffic immune system" is proposed, inspired by the human immune system, to achieve a vision of "zero deaths" in traffic [10] - Principles derived from urban immunology can guide the deep collaboration between people, vehicles, and cities to enhance the resilience and sustainability of transportation systems [10] Group 5: Testing and Standardization Challenges - The lack of unified safety standards, liability frameworks, and testing protocols for autonomous vehicles hinders cross-regional deployment and investment [9] - Existing infrastructure, such as road and communication networks, is insufficient to support large-scale autonomous vehicle operations, particularly in rural areas [9] - The need for real-time interaction between autonomous vehicles and urban traffic management systems is highlighted, along with the current immaturity of V2X communication protocols [9]

"'零交通事故'是自动驾驶系统不断成熟要达到的终极目标，要实现这一目标，可借鉴汽车排放最终实 现'近零排放'的演进路径。"中国工程院院士、清华大学教授李骏解释，汽车排放"近零"目标是在测试设 备越来越精准完善的引领下实现的，相类似地，预期功能安全测试能力越来越强也将引领自动驾驶系统 向"零事故"的方向发展。 安全测试的要点与环境测试不同。清华大学车辆与运载学院副研究员王红告诉科技日报记者，一些自动 驾驶系统白天工作正常，晚上却出现问题，还有一些在极端场景中"宕机"，都是由于在训练的过程中缺 乏必要的环境要素测试和校正。 我国智能网联汽车发展迅猛，但由智能驾驶系统导致的交通安全事故也引发关注，暴露出汽车智能化、 网联化发展过程中的安全"痛点"。 9月12日，工业和信息化部等八部门印发《汽车行业稳增长工作方案（2025—2026年）》，明确"有条件 批准L3级车型生产准入"。 如何以安全领域的技术创新推动我国汽车产业智能网联技术的产业化应用？9月13日—14日，2025世界 智能安全大会在陕西西安举办，百余名全球科技领域顶尖专家、多名中外院士围绕自动驾驶技术发展趋 势和安全挑战展开深度研讨。与会专家认为，L3级车 ...