On August 26, I attended the first Ideal Safety Academy. At first, I thought this was just another dynamic experience event before the launch of the Ideal L9. But after a day of experience, I found that this “safety academy” was different from what I imagined.
Lesson One—Theory
The first speaker was the Ideal product director, Zhang Xiao, who explained how Ideal created a “safe car” with the L9.
The content of the whole lecture included many technical details. In summary, the Ideal L9 first started with the design of the car body structure, which can absorb the impact of collisions to the greatest extent possible to ensure the safety of passengers inside the car.
We all know that AEB is a very important function for passengers in the vehicle in an emergency situation. The Ideal L9 spent a lot of time testing and adjusting the AEB function to enable the vehicle to intervene in dangerous situations at the first possible moment and avoid accidents.
Finally, Zhang Xiao pointed out the many common misconceptions among owners of intelligent electric vehicles in their usage. The most common one is that after knowing the vehicle’s intelligent functions at a superficial level, they completely trust the vehicle to handle all driving scenarios, but in reality, this can be very dangerous.
After discussing the basic safety functions of the vehicle, the next topic was the part I was most interested in—the intelligent driving. The speaker who came up to the stage was Ideal’s intelligent driving product manager, Zhang Zhelan.
Zhang Zhelan first introduced us to the latest intelligent driving system equipped on the Ideal L9—Ideal AD MAX. This system not only has powerful intelligent sensing and processing hardware but also has the latest in-house developed algorithms from Ideal.
From the previous supplier solution to the current self-developed solution, the ideal is constantly optimizing the entire closed-loop development process of intelligent driving. It not only greatly improves efficiency but also allows for the fastest improvement of areas that were not well done in the past through user data. In the era of the explosive growth of intelligent driving, a more efficient development process is undoubtedly an ideal “sharp sword” in terms of intelligentization.
From the Ideal ONE without a lidar to the L9 with a lidar, how much has the perception capability improved was introduced using a very intuitive graph in the classroom. Simply put, the function of the lidar is to cover scenes that cannot be perceived by the camera, enhancing the perception capability of the vehicle. However, as can also be seen from the graph, there is still a part of the dashed circle that is not covered by the two circles. In other words, even with a lidar, there are still scenes that vehicles cannot perceive at the current stage.
When talking about the AEB function of the Ideal L9, I learned that when setting up the Ideal, they also considered two special scenarios, “passengers not wearing seat belts” and “adults carrying children”. Since triggering the vehicle’s active braking in these two situations could actually cause greater harm to passengers, the vehicle will disable the AEB functionality, and the driver will also receive corresponding prompts.
Later, ZheLun introduced us to various safety testing scenarios that we were going to experience on the L9, including “kid ghost head” and “disappearing front car”, which are scenarios we may encounter in daily driving.
Finally, ZheLun emphasized that the intelligent assisted driving of the Ideal L9 is tailored for home users. However, intelligent assisted driving does not equal autonomous driving.
At the end of this class, ideal assigned us to different experiential groups, and each group experienced the dynamic experiential part in a different order. At this point, after sitting at the desk for over an hour, I am a little impatient.
Lesson Two – Experience
What was your favorite subject when you were in school? I believe that many people’s answer would be physical education, because PE class can give us a sense of freedom. Ideal’s Lesson Two gave me the same feeling.
Today’s experience took place at the Beijing Kai Ze Automotive Sport Experience Center. I heard that this place is being regularly rented by Mercedes-Benz AMG. When I first stepped into this place in the morning, I even thought we were going to drive the Ideal L9 on the racetrack. As a matter of fact, we did “get off the racetrack” today.
However, our task for today is not to run the “fastest lap time for a full-size SUV” with the Ideal L9, but to experience its various safety features on the arranged track.
According to the earlier arrangement, I was assigned to Group B. The first experience project was urban road driving.
Urban Road Driving Experience
The urban road driving experience is divided into two parts. The first part consists of three projects:
- Speed bump
- Wet road
- Low grip cornering
Before we started, the head coach gave us a brief introduction. Afterwards, everyone got into their cars. The coaches drove a lap first for demonstration, and then it was our turn to drive and experience it ourselves.
Driving over speed bumps is very common in cities and is the least challenging scenario. The suspension of the Ideal L9 is a combination of front double wishbone and rear five-link, plus air suspension and CDC. Even when passing through two speed bumps continuously at a speed of 50-60 km/h, the suspension is very clean and sharp, and we only feel a slight vibration in the car with no aftershocks.
The slippery road surface project is quite challenging, simulating the situation when vehicles slip while turning in rainy weather. The sprinkler is set up in the middle of the field and continuously sprays water to keep the road surface slippery. Our goal is to drive in circles on the wet and slippery road around the sprinkler, also known as “drifting“.
Before starting, we need to open the “slippery road surface” mode. During the coach’s instruction, I experienced it while sitting in the passenger seat. As the vehicle was driving counterclockwise, the right side of the vehicle was on the outer circle of the circle. At that time, I felt like someone was pressing me against the car door. But the vehicle did not spin out, obviously the vehicle’s ESP intervened.
When it was my turn to drive, obviously I was not as skilled as the coach. I still kept the throttle down when I was driving. I clearly felt that when the car was about to start to slip, the ESP intervened to help stabilize the vehicle, but it didn’t directly cut off the power output, so I didn’t feel the power instantly drop. Some students drove too fast and the vehicle ran out of the original circle, but even so, there was no spinning out of control.
The third project, low grip line change, is also interesting, a somewhat special “moose test”. We need to make two consecutive quick lane changes in the prepared route. The average speed throughout the process is 50-60 km/h. The road section between the two lane changes is not an asphalt paved road, but a concrete road with low adhesion.
The coach’s demonstration made me feel like I was doing a moose test. Although I was “pressed against the door panel” again, it was not difficult to pass through at this speed without touching the cones.
The second half of the urban road driving experience is Golden Cone Challenge. However, compared to the regular Golden Cone Challenge, this one is more like a “driving test”.
The entire test process is as follows: after starting from the starting point, you must first pass through an “S” bend, then enter a dead end composed of water horses, turn around and come out. You can only perform one reverse maneuver, which is also the most difficult part of the whole test. After coming out, there is a long-distance reverse driving, followed by the obstacle avoidance of manhole covers, both of which rely on the 360-degree camera system. Finally, it is simulating the conventional parking and reversing of a vehicle into a parking space. After completion, drive towards the finish line and park the vehicle in the blue cone area.
Due to spending too much time turning around in narrow areas when I was driving, I was criticized by the instructor for taking a “driving school exam” after reaching the destination. But overall, with the help of 360-degree panoramic imaging and transparent ground function, it’s not difficult for the Ideal L9, even as a full-size SUV with a length of over 5.2 meters, to pass the above test. The only thing that takes a little time is turning around in a dead-end, but in reality, there is no limitation to only being able to reverse once.
Intelligent Driving Experience
There are 4 parts to the intelligent driving experience:
- Ghost Child Detection
- Disappearing Front Vehicle
- Rollover Vehicle Identification
- Water Horse Identification
Ghost Child Detection can be said to be a common dangerous situation on the road. I believe everyone can see some heartbreaking tragedies on various video platforms from time to time caused by parents not holding onto their children, resulting in the child being hit by a car while crossing the road.
The setting for Ideal this time is that a van is parked to the front right of the road, and an Ideal L9 is parked behind it. Another Ideal L9 with an antenna on top drives from the left at a speed of 60-70 km/h. When it approaches a certain distance, the top antenna of the vehicle emits a signal, and the antenna of the dummy device receives the signal and then crosses the bread van’s head, simulating the ghost detection scenario.
Firstly, since the time when the signal is received by the dummy is different every time, the appearing time of the dummy is different each time. When I was experiencing it, the dummy appeared when the vehicle reached the tail of the parked L9, so the distance that the vehicle recognized the ghost detection was about 15 meters, which was quite sufficient. Then the AEB function of the vehicle was triggered until it was completely stopped, and there were several meters of distance from the dummy.
However, in subsequent tests, sometimes the dummy only appears when the vehicle reaches the head of the L9, and the stopping distance of the vehicle at this time seems not so “safe,” but it can also ensure that a collision with the dummy will not occur. The Ideal staff secretly told me that the price of this dummy equipment is quite high.
The second project, Disappearing Front Vehicle, refers to the situation when the vehicle follows another car normally, and the front vehicle quickly changes lanes to other lanes due to obstacles on the road, so it is called the “disappearing front vehicle.”
I had encountered such a situation myself before. It was drizzling, and the two cars in front suddenly swerved to the left and right lanes. It was only then that I saw a white Tesla parked on the lane ahead.
Since I had confirmed that the car behind me on the right had a safe distance, I quickly swerved into the right lane to avoid the stationary car. When I checked the dashcam afterwards, I found that not only was the car stationary, but the hazard lights and brake lights were not even on. Despite my slow speed at the time, it would have been very dangerous if I had collided with it.
The test involves two cars, one in front of the other, accelerating to 70 km/h. The rear car then engages the Lane Keeping System (LCC) to maintain speed. When the distance between the rear car and the stationary model is about 30 meters, the front car will swerve to avoid the obstacle, leaving the rear car to brake on its own.
The difficulty of the test is the short reaction time given to the rear car, which tests the perception and recognition abilities of the intelligent driving system. As a member of the first batch of testers, I ran the test twice, and the assistance function of the vehicle could brake and stop both times, without triggering the Automatic Emergency Braking (AEB) function.
What do I mean by assistance braking? When braking in the assistance mode, the vehicle brakes heavily at first, but as the speed decreases, the vehicle gradually comes to a stop.
But if the AEB function is triggered, it will be like the driver slamming on the brakes with full force, effectively stopping the car in an instant. Most of the subsequent test vehicles triggered the AEB function, and the main reason was that repeated heavy braking caused overheating of the brake pads, which in turn reduced braking force. At this point, the assistance braking is insufficient, and the AEB must intervene to avoid a collision with the obstacle.
In the most extreme case, the rear car failed to decelerate for about 1 second after the front car left the vision. Although the AEB system intervened later, it was still too short a distance to stop the vehicle, which collided with the obstacle.
The rollover vehicle test is similar to the water spray recognition test. The vehicle is first accelerated to 80 km/h, and then the Lane Keeping function is activated to automatically detect obstacles and stop the vehicle.
For the L9, which has a LIDAR system, both of these tests are relatively easy. From the vehicle visualizations, the system was able to detect the obstacle from over 50 meters away and began to brake slowly. The vehicle ultimately stopped about 5 meters away from the obstacle.
It is worth mentioning that a real scrapped car is used for the rollover test. If it’s crashed like the model just now, it’s not as simple as rearranging the model for testing.
Special Terrain Driving Experience
Finally, we come to the special terrain driving experience session. This session is actually about taking the LI L9 off-road. When I first entered the site, I was quite shocked. Isn’t this a test project that only off-road vehicles can demonstrate? Does anyone really buy the LI L9 as a family car and use it as an off-road vehicle?
Later, I understood that LI did this not because they really wanted the car owners to use this car for off-road driving, but to tell everyone that if they really encounter such extreme road conditions, the LI L9 can also “go wild“.
There are 5 scenes in the special terrain driving experience:
- Side slope driving
- Small steep slope descent
- Water wading section
- Bomb crater road section
- Large steep slope descent
Friends who often participate in off-road driving experience activities should be familiar with these tests, but how will a family SUV perform on these road sections?
First, we drove the LI L9 up a slope with a gradient of 30-40 degrees. As the co-pilot, I experienced the feeling that “the car is going to roll over” for the first time. Although the vehicle leaned heavily, it was not difficult for the vehicle to pass this slope. I guess it is due to the battery pack at the bottom of the vehicle that increases the weight of the entire vehicle, which makes the vehicle stick to the slope steadily during driving.
Next, we came to a small steep slope descent. The “steep slope descent” function needs to be turned on before climbing. The slope angle is still quite large, estimated to be about 40 degrees, so the coach repeatedly emphasized that the foot must not be released from the accelerator pedal when climbing. At this time, the LI L9’s intelligent four-wheel drive system came into play. Even climbing such a steep slope was not difficult. Although the scene during the descent was quite daunting, the vehicle smoothly came down.
The third item is about water crossings, which I also encountered once in front of my door. At that time, I saw a large area of standing water on the road and did not dare to drive through it, because the challenge in water crossings is not just that drivers do not know how deep the water is or whether there are obstacles under the water, and that day I waited until a BYD Qin drove past without any incident before I followed suit.
In this test, there is a 50 cm deep water pool set up in the arena, which is actually deeper than most standing water scenarios we encounter in daily driving. However, the instructor told me that the depth of this water pool is actually 60 cm.
Before entering the water, the range extender mode needs to be set to “pure electric priority” to force the range extender to stop working. In the water, the Ideal ONE L9 is like a big ship sailing with a car full of people. However, I think the most difficult part of the whole process is not driving in the water, but the process of entering and leaving the pool. The ramp during both entry and exit is very steep, steeper than most underground car park entrances and exits, and knocking and bumping is a common occurrence during the second round of driving when everyone tries it themselves.
The next item, driving through a shell pit, is probably the most classic test item for SUVs because it requires functions such as differential locks that only SUVs have. However, the Ideal ONE L9 tells us that it is possible to pass through the shell pit with just four-wheel drive alone.
Before starting, you need to enable the “off-road rescue” function. While driving, when the system recognizes that one wheel of the front or rear wheels is slipping, it will actively enhance the power of the other side to help the vehicle move smoothly. Coupled with air suspension and CDC’s help, although I still felt quite a bit of ups and downs inside the car, it was less bumpy than I expected, thanks to the Ideal ONE L9’s suspension system.
Lastly, the climax of this experience: steep slope descent. Compared to the previous minor slope descent, this steep slope descent can only be seen as “appetizers”. If I had to summarize the entire experience in one sentence, it would be that I sat in a roller coaster called Ideal ONE L9.
Similarly, this function requires enabling the “steep slope descent” function in advance, and when going uphill, having had the experience of the previous minor slope descent gave me more confidence. And the Ideal ONE L9 also proved that even with such a long and steep slope, I can still climb up.
But things are different when it comes to downhill. With the huge noise made by the vehicle, apart from the instructor sitting in the driver’s seat, I believe everyone on board has a little bit of fear. Fortunately, even though the sound was loud, we landed safely.
After the first round of driving led by the instructor, media trainees can personally sit in the driver’s seat and experience another round. However, I am not confident enough in my own driving skills to try off-roading with the Ideal L9, so maybe I will regret giving up this opportunity in a few years.
Lesson Three – Q&A
When we finished experiencing all the test projects and returned to the classroom from the field, the classroom had been set up like a “big forum.” As the saying goes, the more you see, the more questions you have. When everyone sat back in the classroom, I think most people, like me, still have various questions about the Ideal L9.
“Today, we did not experience all these extreme avoidance, extreme accident scenarios, and extreme road conditions in one day just to show how awesome we are. It is to tell everyone through this way, where is the boundary of our vehicle ability. It’s dangerous when it exceeds this boundary, but if we use it well within the boundary, it can help us avoid some accidents and improve driving safety.”
Before formally starting the Q&A session, Zhang Xiao first summarized today’s experience.
After that, the questions raised by everyone, related to the AEB function of the Ideal L9, were the most, including the triggering conditions of AEB, the perception hardware involved in AEB recognition, the boundary of objects recognized by AEB, and whether AEB participates in deceleration during intelligent driving, etc. ZheLun and Zhang Xiao answered them one by one.
I also raised several questions. First, I am curious about the size limit of objects recognized by AEB. For example, if I set up a tripod on the road, or encounter a box on the road, can the Ideal L9 recognize it and stop?
But ZheLun told me that my idea was wrong from the beginning. The condition for the Ideal L9 to recognize and trigger the AEB function is not how big an object in the vehicle’s travel path is, but if it is a “recognized object”.
For example, when the vehicle recognizes objects such as four-wheeled vehicles, two-wheeled vehicles, pedestrians, overturned vehicles, and water troughs in the traveling path, and the driver does not brake in time, the AEB actively brakes because the vehicle believes that these objects hinder normal driving.## Translation
As an example, when we are driving on our own, if there are objects like plastic bags or leaves in front of the vehicle, we won’t emergency brake or change lanes quickly because we know that hitting them won’t have any impact on the vehicle.
However, if we see a brick or a wooden board, we will definitely try to brake or avoid them because we know that hitting or running over them could likely cause damage to the vehicle.
At this point, you should have realized that the key to whether a vehicle triggers AEB is the vehicle’s “perception”. The objects trained in the database model are what the vehicle recognizes. Anything that is not in the database model is not recognized by the vehicle. Faced with something that the vehicle doesn’t recognize, the ideal choice is not to let the vehicle activate the automatic braking.
Why? Because in the eyes of engineers, “false recognition rate” is actually more important than “recognition rate”. Imagine if the AEB function is triggered while driving at high speed, it would be so dangerous, so the boundary for triggering AEB is very important.
Now, let’s answer the question earlier: Why doesn’t the vehicle brake automatically when there is a box on the road? It’s because, in the eyes of the vehicle, this is just a medium-sized rectangular object, and the vehicle has no perception of the material or other properties of the object. If we add the rectangular model to the database model, then all vehicles will recognize and stop for objects of this size on the road, which is actually unsafe. AEB’s main working scenarios currently are to avoid extreme traffic participants or scenarios that endanger traffic participants.
Finally, both of them also stated that the various safety functions of the L9 will be continuously iterated through OTA updates, including adding lidar data fusion to AEB and adding more objects to the vehicle data model library. The delivery of the vehicle is just the beginning.
For the Ideal Safety School, it is also just a beginning, and Ideal indicates that such activities will continue in the future.
Conclusion
The Ideal Safety School ended after a day, but my heart was not calm for a long time even after returning home.
What is safety? How to ensure safe driving? I believe everyone has a different answer. Then let me ask again, have you really ensured safe driving at all times? I cannot give a definite answer on my own.
What has the Ideal L9 given us? It is to deliver a vehicle with various auxiliary safety functions to our hands, which helps to avoid dangerous situations that may occur due to a driver’s temporary lack of attention. At the same time, Ideal tells us, “don’t completely trust it”.
“I hope all media teachers can spread the word that assisted driving is not autonomous driving after experiencing it.”
After experiencing the rollover vehicle and water horse recognition project, the coach said this sentence to us, and this sentence is the most impressive one for me that day. Undoubtedly, the keyword “safety” runs through the whole activity, but in my eyes, the keyword “assistance” is also running through the whole activity.
The reason why assistance is assistance is that the vehicle helps the people in the car within its capability to ensure their safety. But if the whole driving process and extreme situations are handed over to the vehicle for processing, it means that the driver has given up “safety” and the responsibility for the safety of the people in the car. At this time, the vehicle cannot make decisions for the driver.
It seems that since 2020, some car companies have included “autonomous driving” in their marketing language, which has caused many people to mistakenly believe that after buying this car, they can let the car drive itself and the driver can do whatever they want. Now, Ideal is the first to come forward and emphasize that it is not autonomous driving, but assisted driving. I fully agree with this approach and hope that all car companies can follow suit.
After seeing so many tragedies caused by the incorrect use of assisted driving, I hope those who read this article can remain clear-headed and engrave “safe driving” into their hearts.
This article is a translation by ChatGPT of a Chinese report from 42HOW. If you have any questions about it, please email bd@42how.com.