A rational view of the design of AEB automatic emergency braking system

With the two major traffic pillars of Xiaopeng and Huawei, and the participation of many organizations such as DoCar, AEB (Autonomous Emergency Braking) has become the focus of everyone's conversation. Many users even think that this is an intelligent driving function that can automatically brake in any emergency situation. Objectively speaking, this understanding of consumers may lead to serious traffic accidents. After the traffic, we hope to provide you with some opinions and ideas!

Key points about Automatic Emergency Braking (AEB):

1) Emergency braking is not triggered at any speed. There are differences in thresholds between standard regulations and tests by various car companies, so the upper limit of the AEB function on each car is different, and the speed range for the function to be activated is different. AEB is just a helper, and it is not easy to figure it out in actual combat!

2) The existing AEB technology has a weak perception of ultra-low speed and stationary obstacles. The driver needs to pay attention to the road ahead at all times. Emergency braking is the last line of defense and cannot be relied on for daily driving! It is more dangerous to deal with it as soon as possible and save it for AEB.

3) AEB technology has limitations on the types of obstacles it can detect for vehicles in different states. It can identify common obstacles such as the vehicle directly behind, two-wheeled vehicles and pedestrians, but has limited ability to trigger special-shaped obstacles such as wheelchairs, cones, and sideways parked cars. There is also the technical aspect of AEB. Whether a more advanced software form is used, especially whether smart driving is turned on or not, many states are different. This is a selling point, but it will take time for users to understand.

4) Of course, the most important thing is that if you suddenly step on the accelerator and guess wrong, the AEB function of this car will not be triggered. Although the smart car detects the obstacle, it thinks that you have active driving intentions. Can you do something fancy? At this time, the AEB function will not be called out, and you will take the blame!

The reason why the AEB performance of existing mass-produced models varies greatly is mainly due to the differences in perception sensors and decision-making computing power. For the AEB function, our expectation is to meet the challenge of making accurate braking actions at the right time. The core demand is that the AEB false alarm rate is required to be no more than once within 100,000 kilometers. It should neither be triggered too frequently nor not triggered at all. The difficulty is not low!

The vehicle's perception capability must first identify the target type, and the requirements for accurate identification of the target type are extremely high. AEB has extremely strict requirements on vehicle perception capabilities to avoid false alarms. Due to the different sensor configurations of existing mass-produced models, single sensor solutions such as vision or millimeter-wave radar are limited by sensor performance, resulting in different upper limits of AEB function performance. The AEB function of multi-sensor fusion perception has made significant progress, especially the AEB function involving lidar has performed better. Lidar enables the AEB function to detect obstacles more richly and improve the system's accurate recognition rate of obstacles, thereby further improving the AEB function.

Of course, in the long run, the core of intelligent driving still has a global perspective, and can choose the right path without braking to avoid this scenario. Those who brake casually are not experienced drivers. A gentleman does not stand on a dangerous wall. With appropriate path planning, stay away from danger and hidden dangers. This is what we hope to do best!

The most important thing is that as AI is introduced faster and more effectively in smart cars, the single function of AEB will become a thing of the past! We will conduct comprehensive tracking and discussion around smart driving!