Abstract:
The National Standard Information Public Service Platform shows that the national standard "Safety Requirements for Autonomous Driving Systems of Intelligent Connected Vehicles" (GB 44721-2026) has been released to the public. This is also my country's first mandatory national standard for the safety requirements of L3/L4 intelligent networked vehicle autonomous driving systems. It has been officially finalized and will be implemented from July 1, 2027. All models on sale equipped with high-end autonomous driving functions must subsequently meet the rigid requirements of this standard.
This national standard clearly delineates the general technical requirements for autonomous driving system dynamic driving task execution, human-computer interaction, user notification and other dimensions. It also clearly marks the safety assurance obligations that vehicle manufacturers need to implement, as well as the test access conditions for corresponding functions.
For L3 level autonomous driving functions, the standard specifically clarifies the technical redline for takeover capability monitoring, and refines three hard requirements for seat belt status monitoring, driver presence monitoring, and pre-takeover dynamic driving task execution capability monitoring, to prevent users from misjudging the autonomous driving status and causing accidents.
For L4 level autonomous driving functions, the standard also separately sets up exclusive technical terms, providing clear implementation guidelines for dynamic driving task execution logic, background remote assistance rules, and human-computer interaction logic in special scenarios.
Previously, the relevant person in charge of the Ministry of Industry and Information Technology publicly stated that this standard is completely based on the current status of my country's industrial development and the actual needs of road supervision, taking into account the technical feasibility of the current industry, the industrial adaptability of different car companies, and also fully considering the practical attributes of the implementation link, and finally builds an autonomous driving safety requirement system with complete dimensions, clear requirements, and fully adapted to the national conditions of domestic roads.
The first is to improve the enterprise's full life cycle safety assurance mechanism and build a solid foundation for the safety of high-level autonomous driving. The standard requires vehicle manufacturers to build safety assurance capabilities covering the entire life cycle of product design and development, production and manufacturing, and post-delivery from four core dimensions: safety policy formulation, full-process risk management, safety capability verification, and dynamic iterative optimization, so as to reduce the potential safety risks of autonomous driving systems from the source.
At the same time, the standard also clearly requires that during the development and verification phase of autonomous driving systems, car companies must complete simulation tests, closed site tests, and open road tests of corresponding magnitudes in accordance with specifications to ensure that relevant test resources and verification conditions fully meet the safety verification requirements before product launch.
The second is to strengthen the system's dynamic driving task execution capabilities and ensure the safety of ordinary users from the bottom up. The standard requires that the overall safety level of the implemented autonomous driving system must be at least the same as that of a qualified and fully focused human driver. During the entire process of performing dynamic driving tasks, it must not cause any unreasonable additional safety risks to users in the car and other road participants.
On this basis, the standard also clarifies the safety boundaries for the system to perform dynamic driving tasks in various extreme scenarios, and also provides clear normative guidance on the conditions and execution processes for vehicles to trigger the minimum risk strategy.
The third is to comprehensively standardize human-computer interaction and user notification requirements to prevent the risk of misuse and abuse of autonomous driving from a rule level.
The standard requires that there should be no security loopholes in the entire process of activation and exit of the autonomous driving system. It also clarifies the prompt signal specifications for different states such as system readiness, normal activation, and imminent exit. All prompt information should not be too professional and obscure, and must be understandable by ordinary drivers at a glance.
For L3 autonomous driving systems, the standard clearly requires that it must be equipped with a continuous monitoring function of the driver's ability to take over, to prevent users from turning on high-level autonomous driving even when they are unable to take over.
At the same time, car companies are required to clearly inform users of the driving automation level of the vehicle, the range of scenarios that the function can cover, known functional limitations, control strategies in special scenarios, and user role transition rules in different states through multiple public channels such as official websites and in-vehicle central control terminals, in addition to the product manuals that come with the vehicle, so as to guide ordinary users to establish a correct understanding of high-level autonomous driving and avoid accidents caused by over-reliance on functions.
The last step is to improve the multi-dimensional inspection and testing methods to ensure that the standards can be truly implemented and not distorted from the implementation level. The standard has established a complete inspection and testing system that includes enterprise safety assurance capabilities verification, full-process safety file verification, and product confirmatory testing. It conducts comprehensive verification on core dimensions such as the full-link safety assurance capabilities of car companies and the integrity and robustness of product safety files. It will block high-end autonomous driving models that do not meet safety requirements from the access stage.
The formal implementation of this mandatory national standard is equivalent to delineating a unified safety bottom line for the domestic high-end autonomous driving industry. In the past, various vague propaganda and the chaos of secretly substituting the concept of intelligent driving levels that many car companies have defined themselves will be directly prohibited. The entry threshold for the entire autonomous driving industry will be significantly raised. When ordinary consumers subsequently purchase models equipped with high-end intelligent driving functions, their rights and interests will also be endorsed by clear national standards, and they will no longer have to pay for the safety risks caused by the industry's irregular involution.
Comments