The AEB function of Huawei's Yu Chengdong and Xiaopeng Motors' He Xiaopeng, who previously engaged in aerial confrontation, will now become a standard feature in automobiles. Recently, the "Technical Requirements and Test Methods for Automatic Emergency Braking Systems of Light Vehicles" (hereinafter referred to as the "new national standard") has been drafted and entered the stage of soliciting public opinions. This standard will replace the current national standard GB/T 39901-2021 and become the latest national mandatory standard for the performance and testing methods of automatic emergency braking systems (hereinafter referred to as AEB systems) for passenger cars. This policy signal indicates that automotive safety technology will undergo a deep transformation from "functional stacking" to "safety effectiveness".

The problems exposed by the AEB system at the current stage

Nowadays, with the rapid popularization of assisted driving technology, AEBs are being installed on an increasing number of vehicle models. According to data from the China Association of Automobile Manufacturers, the overall installation rate of passenger cars reached 56.5% in the first two months of this year, and the AEB installation rate in the new energy passenger car market has reached 62.9%. However, behind the rapidly growing loading rate, AEB function failures, false triggers, and other situations frequently occur due to software, hardware, and algorithm issues, turning the original safety technology into a "safety hazard".

According to data from Chezhiwang, from 2019 to May 2025, the cumulative number of complaints about AEB malfunctions exceeded 650, covering many car brands, among which new force brands accounted for a large proportion, such as Tesla, Xiaopeng, Leapmotor, Ideal, and Wenjie, all of which were affected. Based on complaints from the Car Quality Network and recent traffic accidents caused by AEB malfunctions, there are several reasons why AEB malfunctions.

One is the failure of AEB function. This includes two types of situations, complete failure and functional triggering "lag". The former is easy to understand, that is, the AEB function was not triggered, mainly due to hardware or software problems in the system, such as sensors failing to recognize obstacles ahead, software algorithm bugs, etc. Recently, in the high-speed collision and fire incident of Xiaomi SU7, the AEB function of the car was not triggered. The official explanation given is that the AEB function equipped on the vehicle currently does not respond to obstacles such as cones, water horses, stones, and animals. The latter refers to the fact that the AEB function has not been triggered, but the triggering time is delayed compared to the normal state, resulting in insufficient execution time of the braking system, and the vehicle cannot slow down or stop in time.

The second type is the accidental triggering of AEB function. This problem is quite common in daily life, referring to the AEB function being triggered when there is no collision risk ahead. This has a significant impact on the comfort and safety of vehicle operation. Especially in terms of safety, many rear end collisions are caused by the front car braking abruptly. Compared to AEB function failure, the probability of accidental triggering is much higher. It can be seen from the complaint data of Chezhiwang that complaints related to AEB false triggering account for about 47.4% of the total. According to Mr. Chen from Suzhou, when there were no obstacles in front of the vehicle and the speed was around 50km/h, the AEB emergency brake was triggered abnormally, almost causing multiple rear end collisions.

From a working principle perspective, the triggering of AEB requires three steps: perception, decision-making, and execution, with perception being the most important. At the perceptual level, the technological routes of different enterprises are also different. Although they are all called AEBs, the functions they implement vary greatly. For example, the AEB system equipped with only 77GHz millimeter wave radar is the most basic technical solution and the lowest cost. It has a wide range of applications, but its functions are easily affected by external environmental interference, leading to failure. There are also pure visual solutions that rely on cameras for perception. The drawback of this solution is that it has limitations in recognizing stationary objects and is also susceptible to environmental interference such as lighting. The currently effective solution is multi-sensor fusion, which combines millimeter wave radar, laser radar, and cameras to avoid their respective shortcomings and ensure more accurate recognition. But this solution requires high performance from the processor, so the cost naturally increases.

AEB mandatory national standard arrow on the string

In fact, with the gradual implementation of advanced assisted driving functions such as high-speed navigation and city navigation, the requirements for the use of AEBs have become higher, from recognizing vehicles ahead to recognizing pedestrians, irregular obstacles, small objects, and even small animals. The launch of the "new national standard" can be said to be timely, not only changing the AEB function from "optional" to "standard", but also putting forward stricter requirements for its performance and testing methods.
AEB has become a national mandatory standard for assisted driving, and it should strictly adhere to the safety bottom line

1. Expanded scope of application. Compared with the current recommended national standard, the "new national standard" has expanded the scope of application of the standard, not only M1 class vehicles (passenger cars) need to meet the national standard requirements, but N1 class vehicles (low-speed cargo vehicles) have also become applicable objects, which will make the AEB system enter a more comprehensive and mandatory popularization stage.

2. Upgrade testing methods and standards. Firstly, the "new national standard" has added an assessment of the recognition ability of vulnerable traffic participants such as pedestrians, bicycles, and scooter type two wheelers, requiring the AEB system to effectively recognize pedestrians and two wheelers within the range of 20km/h-60km/h. At the same time, night and low visibility environment testing has become mandatory, and the system's false triggering rate must be less than 0.1% (the current standard is 1%). In addition, the current national standard focuses more on testing vehicles in straight road environments, while the new national standard requires testing in complex scenarios such as intersections, right turn following, and stationary bicycles in the opposite direction.

Secondly, the "new national standard" has added simulation testing items and requires that at least 30% of all tests be conducted through on-site testing, reflecting the more detailed and strict requirements of the "new national standard". Among them, it is required that the pass rate of vehicle to vehicle tests should not be less than 90%, and the pass rate of vehicle to pedestrian/bicycle tests should not be less than 80%. This standard is not only higher than the current recommended national standards, but also in line with the EU U-R152 regulation.

In addition, the "new national standard" has raised the upper and lower limits of AEB triggering speed. The current national standard requires the minimum triggering speed of AEB system for M1 class vehicles (passenger cars) to be ≥ 15km/h. In practical application, most industries set the effective speed range of AEB system at 15-60km/h. The new national standard requires that the AEB system of M1 class vehicles (passenger cars) must be activated when it detects a collision risk with the preceding vehicle within the speed range of 10-80km/h; N1 type vehicles (low-speed cargo vehicles) are activated within the speed range of 10-60km/h.

It can be foreseen that after the implementation of the "new national standard", AEB function will no longer be a marketing selling point for car companies, but an essential technology for active safety, which can be regarded as a watershed in the history of Chinese automotive safety. However, it should be noted that the "new national standard" only sets the AEB admission threshold for new cars, which can be considered as a "passing line". For AEB systems, there is still a significant difference between 'usable' and 'easy to use'.

On the one hand, car companies need to upgrade their technology by designing sensing solutions and algorithms for three usage scenarios: urban road conditions, highway conditions, and pedestrian protection. This will improve the sensing accuracy of sensors, increase the upper limit of AEB capabilities, and achieve more stable triggering. At the same time, when formulating the timing and strategy for AEB intervention, it is not enough to simply consider reducing the false triggering rate. More variables and testing scenarios should be introduced to make the triggering logic of AEB closer to real driving scenarios. For example, the driver's operation of the vehicle is the most important variable, and the system needs to have stronger judgment ability to confirm whether the driver is distracted during the driving process and has not actively controlled the vehicle, so as to intervene actively.

On the other hand, in terms of standards, it is necessary to align with international standards. Currently, domestic testing standards are not as strict as those abroad in certain complex scenarios. For example, the EU's UN-R152 regulation requires the AEB system to be activated within the speed range of 10-60km/h in scenarios such as vehicle to person and vehicle to bicycle. In addition, the weight of AEB testing can be further increased in domestic collision testing, especially AEB VRU (AEB pedestrian/bicycle). At present, although C-NCAP has separated it from the active safety ADAS test section as an important part of VRU protection in the 2024 version of the testing regulations and released it in the form of an independent appendix, and has also added dangerous scenarios such as intersections and "ghost probes", the weight of related tests in the entire vehicle is still relatively low.

How to maintain the safety bottom line in the competition of assisted driving

As the "last line of defense" for vehicle safety, AEB's core value lies in avoiding collision risks caused by driver reaction delays. With the gradual popularization of assisted driving functions, how to maintain this safety bottom line in market competition requires the joint efforts of the industry, car companies, and consumers.
AEB has become a national mandatory standard for assisted driving, and it should strictly adhere to the safety bottom line

At the industry level, it is necessary to continuously improve relevant technical standards and testing scenarios, such as adding more extreme testing scenarios that are in line with the actual road environment in China in C-NCAP and CIASI evaluations. At the same time, "assisted driving safety rating" can also be introduced, similar to Euro NCAP's "Driver Monitoring (DMS) rating", to promote industry transparency.

For car companies, on the one hand, it is necessary to optimize their own technological routes. For example, some car companies adopt a pure visual route, some use laser radar, and some use multi-sensor fusion routes. However, regardless of which one, continuous optimization must be carried out around driving safety, including hardware upgrades and software iterations, to ensure that consumers have no worries when using assisted driving functions. On the other hand, the most important thing is for car companies to adhere to integrity when promoting assisted driving functions, and avoid exaggerating and misleading consumers. Of course, such situations are unlikely to happen again in the future. Recently, the Ministry of Industry and Information Technology issued a notice clarifying the "system functional boundaries and security response measures, prohibiting exaggeration and false advertising".

At the consumer level, the most important thing is to clarify that the AEB system is not foolproof, it is just an auxiliary safety function. At present, due to the different settings of various car companies, the actual performance of AEB functions will also vary greatly. Do not blindly believe in the company's propaganda, but carefully understand it. For example, some car models can autonomously complete braking after triggering the AEB function, while others require driver intervention to achieve braking. Taking the ideal L8 Max as an example, when the vehicle speed drops to 10km/h, the braking action will not be executed, and the driver needs to intervene.

summarize

The implementation of AEB national mandatory standards is an important milestone in promoting industry progress, but for car companies, it cannot just stop at "compliance". In the competition of assisted driving, safety is the true core competitiveness. Only by evolving AEB from an "option on the configuration list" to a "user trusted security line" can we truly realize the value of technological inclusiveness.