From June 22 to 26, 2026, the 199th session of the World Forum for Harmonization of Vehicle Regulations (WP.29) of the United Nations was held in Geneva. The Global Technical Regulation for Automated Driving Systems (ADS GTR), jointly led by China, the European Union, the United States, Japan, the United Kingdom, and Canada, was officially approved and released. This marks the world's first unified technical regulation specifically targeting automated driving systems.
Prior to this, the global autonomous driving industry had long been in a state of fragmented regulation, with countries advancing R&D based on their own experiences. A single intelligent driving solution entering different markets required multiple rounds of adaptation and repeated verification. Some institutions estimate that for a vehicle model entering the EU, US, and Japanese markets, compliance and adaptation costs could account for 15% to 20% of the total R&D investment. This is precisely where the value of the ADS GTR lies. It provides a globally unified blueprint. While the regulation itself does not mandate execution by individual countries, it offers a translatable technical template. Countries can translate it into their own national regulations, paving the way for mutual recognition of compliant vehicles across multiple nations.
It should be noted that China played a pivotal role in the formulation of this regulation. Since 2018, China has successively served as the Vice-Chair country of the WP.29 Working Party on Automated/Autonomous and Connected Vehicles (GRVA) and the Co-Chair country of the Functional Requirements for Automated Driving (FRAV) working group. The Ministry of Industry and Information Technology (MIIT) organized relevant institutions to participate throughout the entire process, submitting dozens of technical proposals focusing on dynamic driving tasks, human-machine interaction, and test verification methods. The release of this regulation, in a sense, represents an international export of China's years of practice in the intelligent driving industry.
Image source: Official website of the MIIT
Regarding this regulation, there have already been numerous reports online. However, today, Intelligent Driving Frontier will analyze its content to explain what it entails and what changes its emergence will bring to the intelligent driving industry.
Before discussing the ADS GTR, it is necessary to first touch upon the domestic situation. In February 2026, the MIIT released the draft for comments on the "Safety Requirements for Automated Driving Systems of Intelligent Connected Vehicles." On June 17, the draft for approval of this standard began its public notice period, with plans to officially implement it on July 1, 2027.
No Mandate on LiDAR or Cameras, Just Asking If It Is Safe Enough
After the public notice for the "Safety Requirements for Automated Driving Systems of Intelligent Connected Vehicles" began, a rumor circulated online claiming that L3 and L4 autonomous driving mandatorily required the installation of LiDAR. In fact, the "Safety Requirements for Automated Driving Systems of Intelligent Connected Vehicles" only specifies the functions of intelligent driving and makes no statements regarding hardware. The "Safety Requirements for Automated Driving Systems of Intelligent Connected Vehicles" has also been uploaded to the Knowledge Planet of Intelligent Driving Frontier; interested readers can go and check it out.
The ADS GTR actually puts forward a core requirement: the safety level of the automated driving system must meet or exceed that of a competent and attentive human driver. This requirement does not sound very difficult, but it is technically challenging to implement. The system does not need to be perfect or achieve zero accidents, but it must be safer than a normal, cautious human driver. The regulation uses minimum safety requirements to define this boundary.
Of course, the ADS GTR also adheres to the principle of technological neutrality. It does not ask whether you use LiDAR or pure vision, how many TOPS of computing power you have, or how large your model parameters are. It only asks one thing: can the system prove that it is safe enough? This approach of focusing on results rather than processes leaves room for innovation across different technological routes.
In terms of scope of application, the ADS GTR covers Class 1 and Class 2 vehicles, namely passenger cars and light commercial vehicles. It targets automated driving systems capable of continuously performing the entire dynamic driving task (DDT), requiring the system to complete the entire chain of environmental perception, planning and decision-making, and vehicle control. Meanwhile, manufacturers must clearly define the Operational Design Domain (ODD) of the system, explicitly informing users about applicable weather conditions, suitable road types, and the applicable speed range.
Furthermore, the ADS GTR has designed a framework covering the entire product lifecycle. In other words, it is not just a matter of building the car and testing it once. From safety management during the R&D phase, to quality assurance during the production process, to continuous monitoring after the vehicle is on the road, every link has corresponding requirements.
Multi-Pillar Approach: Simulation, Proving Grounds, and Roads, None Can Be Missing
The ADS GTR proposes a validation system called the multi-pillar approach. Simply put, it requires manufacturers to jointly prove the safety of the system through three dimensions: simulation testing, closed-course testing, and public road real-vehicle testing.
There is a technical detail worth noting here. In the past, many enterprises were accustomed to using metrics such as how many hundreds of thousands of kilometers were tested or how many cities were covered to prove they were safe enough. However, the ADS GTR requires a different logic: conducting systematic safety argumentation with a structure of claims, arguments, and evidence. Enterprises need to clearly state under what conditions their system is safe, what methods they used for verification, what the verification results are, and how to provide a fallback if the system fails.
In the ADS GTR, the consistency verification between the simulation toolchain and real-vehicle test results is also listed as a hard requirement. Automakers cannot simply claim that the simulation has run for hundreds of millions of kilometers; they must also prove that there is sufficient correspondence between these simulation scenarios and real-world road scenarios.
The regulation also explicitly prohibits vehicles from using AI for online learning during operation. The reason for this clause is that if an automated driving system changes its own model parameters while driving on the road, its behavior becomes unpredictable, making safety argumentation out of the question. The system can undergo OTA (Over-The-Air) upgrades, but the upgrades must be conducted only after verification is completed in a controlled environment.
In terms of human-machine interaction, the regulation requires that information transmission between the system and the user must be clear and unambiguous. Drivers need to know at all times what state the system is currently in, when takeover is required, and how long the takeover will take. These interaction logics are not just icing on the cake, but necessary components of safety argumentation.
Regarding data recording, the regulation also requires vehicles to be equipped with a Data Storage System for Automated Driving (DSSAD) to record safety-related operational data. These data will play a key role in accident analysis and liability determination.
From Function Demonstration to Safety Argumentation: Is the Industry Logic Being Reconstructed?
The implementation of the ADS GTR, in the short term, clarifies the entry barriers for the industry and frees up R&D resources. However, its more profound significance lies in establishing a global consensus for measuring autonomous driving capabilities, providing a measurable and verifiable definition for good autonomous driving.
Over the past few years, the main battlefield for competition in the autonomous driving industry has been the number of functions and the scenarios covered. Whoever can launch urban NOA (Navigate on Autopilot) earlier, cover more cities, and handle more edge cases will have more advantages and a stronger voice. However, the rule established by the ADS GTR is that no matter how powerful your functions are, you ultimately have to pass the checkpoint of safety argumentation. What enterprises need to submit is a complete set of testable, auditable, and traceable safety dossiers.
This directly changes the logic of enterprises' R&D resource allocation. The regulation requires enterprises to establish an audited safety management system, meaning that safety is no longer just the concern of the algorithm team, but a systematic engineering effort that runs through organizational structure, R&D processes, supply chain management, and after-sales service. The construction and verification of simulation toolchains, the compilation and maintenance of safety dossiers, and post-deployment data monitoring and reporting will all become new fixed cost items.
Looking at industry data, the penetration rate of new vehicles equipped with combined driving assistance systems in China reached 70% in 2026, among which the penetration rate of models equipped with Navigate on Autopilot (NOA) functions exceeded 30%. In December 2025, the first batch of two models equipped with L3-level systems was conditionally approved domestically. When the L2+ penetration rate breaks through 60% and L3 products begin commercial delivery, the regulatory framework shifts from lagging to synchronous, which is a typical signal of structural change.
As mentioned earlier, the domestic autonomous driving standard system is also shifting from recommended to mandatory. The recommended national standard GB/T 44721-2024 released in 2024 has been resubmitted for approval in the form of a mandatory national standard by June 2026. Compared with the ADS GTR, the domestic mandatory national standard puts forward more detailed technical requirements for L3 and L4 systems, and innovatively proposes unified standardized test scenarios. As a mandatory standard targeting domestic L3 and L4 autonomous driving, we have also conducted a detailed interpretation of it previously: Mandatory Standards for L3 and L4 Autonomous Driving Are Coming, Is the Industry Ready?
Of course, the release of the ADS GTR is just the beginning. Unified technical boundaries reduce redundant investments across markets. However, how to solidify and thoroughly implement safety argumentation under a unified regulatory framework may represent a more fundamental shift for enterprises that have become accustomed to proving themselves with mileage and city counts.