芝能-烟烟
2025.07.15 00:45

China's mandatory standards for assisted driving: Is it better to be lenient or strict?

portai
I'm LongbridgeAI, I can summarize articles.
​​

Produced by Zhineng Auto

 

Since April, the penetration rate of L2-level combination auxiliary driving systems in the domestic passenger car market has rapidly increased, and the contradiction between technology implementation and practical use has become more apparent.
 

To curb excessive publicity and ensure user safety, the Ministry of Industry and Information Technology has led the formulation of the mandatory national standard "Safety Requirements for Intelligent Connected Vehicle Combination Driving Assistance Systems," which has been officially established.
 

This mandatory standard will clarify the system function boundaries, driver monitoring requirements, and propose specific technical requirements for system detection capabilities, accident handling mechanisms, etc., indicating that the auxiliary driving industry is entering a stage of standardized development with "safety as the core."

 

Whether this standard should be stricter(encouraging scores above 90 to let leading companies develop) or more lenient(recognizing 60 points is enough to allow more companies to participate) has also become a hot topic, and we are here to discuss it.


 

Part 1

What is in the mandatory standard?

 

Let's first talk about the consensus part

 

◎ DMS system
 

In the L2-level auxiliary driving system, the lateral and longitudinal control of the vehicle is continuously taken over by the system, but the driver remains the responsible party. Real-time monitoring of the driver's state becomes the first line of defense to ensure safe operation.
 

The current mainstream DMS systems mainly rely on infrared cameras to capture parameters such as the driver's eye movement, head posture, eye closure duration, and gaze direction, combined with AI models to determine whether there is distraction, fatigue, or loss of attention.

 

However, in the past period, DMS has faced many problems in product implementation: on the one hand, some car companies, considering "user experience," allow users to turn off the DMS reminder function; on the other hand, some DMS systems have not achieved sufficient response loops, only providing mild prompts and lacking deeper linkage responses.
 

The national standard proposes that once the vehicle detects that the driver is continuously ineffective, such as fatigue or distraction and does not respond to prompts, it must actively enable the "emergency automatic side parking system" to smoothly park the vehicle. From the system integration perspective, DMS must form a(VCU/ICU), braking system(EPB/ESC), and communication module(CAN/Ethernet)real-time closed loop.

 

Once an abnormal state is triggered, the vehicle needs to automatically switch control modes, perform actions such as speed limiting, lane changing, side parking, and parking, and repeatedly notify the user of the current system status through HMI(such as HUD, audio, vibration, etc.).
 

In the future, DMS will no longer be an "auxiliary accessory," but the pre-safety logic core of the entire L2 system. Failure detection and safety takeover will become the "entry threshold" for certification and admission.
 

◎ Transparency of the capability boundaries of auxiliary driving systems and verification of environmental adaptability
 

The biggest misconception of the L2 auxiliary driving system is that users equate it with "autonomous driving," mainly due to excessive packaging, semantic ambiguity in car company publicity, and insufficient boundary information in actual product manuals.

 

The new standard clearly stipulates that vehicles must declare the capability boundaries of the auxiliary driving system, including but not limited to applicable road types(highway/urban expressway/general urban road), infrastructure conditions(whether it relies on high-precision maps, traffic marking clarity), weather environment(rain, snow, fog handling capability), lighting conditions(night recognition), and handling mechanisms in dynamic risk scenarios(sudden stop of the car in front, ghost probe, parallel bicycles, etc.).
 

From the engineering implementation level, this means that the auxiliary driving system must build a set of "boundary perception models" that can determine in real-time during system operation whether the current environment has deviated from the designed operating conditions(ODD, Operational Design Domain).
 

This not only involves the accuracy and recognition coverage of the perception system but also imposes higher requirements on the conservative strategy of system modeling. For example, when the system recognizes that it is currently at the entrance of a curve and the visual signal attenuation is severe (such as strong light reflection, white screen in heavy rain), it needs to trigger a "function degradation" prompt and notify the driver to take over through voice, instruments, etc.

 

The standard also emphasizes that the system must achieve "function consistency verification." This requires that under different software and hardware versions(including before and after OTA upgrades), different environmental variables(such as northern winter and southern rainy season), the system response remains consistent. This requirement will force car manufacturers and Tier 1 suppliers to collaborate deeply and establish a broadertest matrix and multi-scenario redundancy evaluation system.

 

Part 2

What are the main targets of "strict" and "lenient"?

 

The controversy here is the night construction detour test(road closure guidance + temporary passage), which simulates a more complex detour situation.
 

The test road is a two-way four-lane road, and the vehicle was originally driving in the rightmost lane. However, the road ahead is closed by rows of "water horses"(a type of plastic barrier) placed diagonally, not only occupying two lanes in this direction but also extending into part of the opposite lane.

 

To allow the vehicle to continue moving forward, the construction party has reserved a passage about 10 meters wide in the middle of the opposite lane, but this passage is also a temporary path surrounded by water horses and traffic cones.

 

The test starts 200 meters away from the water horses, with the vehicle cruising steadily in auxiliary driving mode. The goal is to detect whether the system can timely discover the road closure information ahead and correctly choose the detour path to enter the temporary passage, while ensuring safe passage without collision.

 

Similarly, the qualified standards include:
 

◎ Successfully completing the detour and avoiding collision;
 

◎ Or if a collision occurs, the vehicle speed is below 10 km/h;
 

◎ Throughout the process, the system should alert the driver in advance, and the response should be consistent and reliable.

 

This basically replicates the accident cases that affect the promotion of auxiliary driving. We see that the entire auxiliary driving draft for comments has a lot of content(we will address them one by one later). Whether it is necessary to strictly follow such extreme cases as a one-vote veto system for mandatory standards is actually a big question.
 

 

If we split it, we can see that 100 points is the goal we pursue, the logic of selecting the best in various extreme situations, and 80 points is the high line. 60 points is the basic entry requirement. From a global perspective, the mandatory standards we are formulating may be the most comprehensive in scenarios and the highest in requirements for auxiliary driving standards.

 


 

Summary

 

In summary, the "strictness and leniency debate" of the L2 auxiliary driving mandatory national standard is centered on the "balance between safety bottom line and industry inclusiveness."
 

Is it to be satisfied with "usable is enough" for broad coverage, or to pursue "safe and reliable" high-quality development? Facing real challenges such as night construction detours, should it be used as a strict filter for risk or as a progressive goal for guiding improvement?
 

The answer will profoundly shape the future landscape of China's auxiliary driving industry. We will continue to follow this topic and welcome you to join the discussion.​​​​

The copyright of this article belongs to the original author/organization.

The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.