I can't stand it anymore", "It's very funny at this time", "Because he thinks we are too good at it", "IQ seems to have dropped by a generation", "The experience is not good enough", "I think this logic is wrong ”, “It’s less efficient than my driving”, these are all remarks made by Xiaopeng Motors Chairman and CEO He Xiaopeng when testing urban NGP. Is Xiaopeng Motors’ urban NGP really that weak?
Obviously the answer is no. Because He Xiaopeng also said:
html On June 27, He Xiaopeng, chairman and CEO of Xpeng Motors, released an internal technical test video of Xpeng P5 (parameters | inquiry) on Weibo. .Compared with the comments section of the trailer on the 24th, many netizens responded positively after watching the feature film on the 27th. What valuable information does this feature film bring us? How is the performance of the urban NGP? Let’s look down with questions.
01 What can we see from the video?
Let’s talk about the conclusion first. The performance of the Xpeng P5 urban NGP is not as “unbearable” as the trailer revealed. At least it will be better than what many keyboard warriors think. After driving on the entire test city road, Da Shijun believes that although the performance of the urban NGP is not as good as that of experienced drivers, it is at least much better than many novice drivers.
Under relatively smooth road conditions, the performance of the NGP is better. Pengcheng City NGP's performance is not much different from that of normal people in terms of detours and turns, and is significantly better than competing products with mainstream driving assistance systems. However, in congestion or low-speed following situations, whether it is the following distance, The detour strategy and lane-changing logic are indeed still far behind those of experienced drivers. Even He Xiaopeng said that he still has many years to go before he can catch up with stronger drivers.
Let’s first take a look at the specific encounters during the entire test process. What kind of road conditions, as well as the response strategies adopted by Xiaopeng NGP, as for the improvement methods, we will talk about it later.
In the entire video of testing NGP, we mainly drove on busy traffic sections in Guangzhou, and the situations we dealt with included.
Specifically, Xpeng P5 Urban NGP can directly "pass" in 180-degree U-turn . It shows that the NGP function accurately grasps the radius of the curve, and there is no situation where the reserved position is too small or too large.
Among them, there were non-motorized vehicles on the inside and outside of the curve, and the NGP system stopped and waited for the non-motorized vehicles to leave before starting again, which was not very efficient.
In the traffic light recognition scenario , the urban NGP function of the Xpeng P5 is quite practical. Even if there are vehicles such as buses that can easily block the driver's sight appearing in front, the traffic light signals can still be accurately identified.
In the lane change and detour scenario , the urban NGP of Xpeng P5 is indeed a bit too "gentle".
When changing lanes on busy urban road sections, the opportunity to change lanes is often fleeting. Xiaopeng Urban NGP was not decisive enough to perform lane changes when there was a short-term vacancy for lane changes. It should be noted that when Xpeng P5 Urban NGP was changing lanes to the right, the system stopped the vehicle in the middle of the road when facing a bus vehicle on the rear right and needed to change lanes to the left, which was not adequate in handling large vehicles. intelligent.
When turning left to on the unprotected road section of , the Xpeng P5 NGP can still follow the trajectory of the car in front and the car next to it and pass smoothly without road markings, which is worthy of praise.
In the case of following at low speed, Xpeng P5's urban NGP handles well except that the following distance is slightly longer. It is worth noting that when the vehicle entered the ramp, it encountered a taxi that did not keep up with the traffic and moved at a low speed. The Xpeng P5 Urban NGP did not change lanes and go around. Wu Xinzhou, vice president of autonomous driving technology of Xpeng Motors, explained that such working conditions do not meet the conditions for the Xpeng P5 urban NGP to change lanes and detour, requiring the vehicle in front to be completely stationary.However, faced with such an explanation, He Xiaopeng and Da Shijun were not convinced.
In addition, when facing vehicles in the next lane that affect the vehicle's driving, Xpeng P5 Urban NGP also adopts a more cautious parking and waiting method, which is still not enough for experienced drivers.
In general, although the overall driving skills of the Xpeng P5 equipped with urban NGP are not as good as those of experienced drivers, it is indeed much better than many novice drivers who are not skilled enough, and it deserves recognition. What is even more worthy of praise is that Xpeng Motors and He Xiaopeng himself have not been stingy in exposing the current problems of urban NGP to the public. Isn't this approach better than those car brands that only brag about themselves but cannot actually stand the test? Do you want it good?
02 What are the differences between the performance of urban NGP and LCC-L measured by Da Shijun on urban roads?
First of all, the biggest improvement point of urban NGP and LCC-L (enhanced version of lane centering) is that it can perform autonomous steering, autonomous lane changes and other functions based on high-precision map navigation. It not only greatly reduces the number of driver takeovers, but also overall The performance is also closer to the "autonomous driving" function that people imagine.
It is worth mentioning that before, Da Shijun experienced the LCC-L (enhanced version of lane centering) function of Xpeng P5. When faced with motor vehicles and non-motor vehicles that slightly encroached on the lane, they often used overly conservative brakes to give way. Way. In this urban NGP internal test video, I saw that when a vehicle faces the same situation, it will adopt a more flexible lateral movement method to overtake. It is indeed much smarter than its previous performance and deserves praise. Of course, this does not rule out version iterative improvements.
03 What areas can be improved?
1. During the U-turn, the three non-motorized vehicles shown in the test video did not directly overlap with the driving route. Faced with this situation, experienced drivers would usually pass at a low speed instead of stopping on the road. The braking method used in this test may cause a rear-end collision with a vehicle that is following too closely behind.
2. In traffic jams, the Xiaopeng P5 followed the car a little further, and it was not aggressive enough to follow the car in front. Whether it is a slight congestion or an extreme congestion situation, the slightly longer following distance of the Xpeng P5 makes it easier for other social vehicles to jam. Of course, if Xiaopeng urban NGP can improve the response to starting with a car, it will not only improve driving efficiency, but also effectively reduce the possibility of being jammed.
3. When evaluating the movements of other vehicles, Xiaopeng City NGP can also distinguish between right-of-way and . For example, during the lane change process, the bus on the rear right happens to need to merge to the left. The best way to respond to the Xpeng P5 that needs to merge to the right is to accelerate and open the distance before merging, rather than turning. The light brake stopped in the middle of the road. Accelerating and widening the gap before merging not only reduces the possibility of accidents, but also reduces the bus driver's suspicion of your next move.
4. In low-speed car following scenarios, for example, if the car in front fails to keep up with the traffic flow, the urban NGP system should take the initiative to change lanes and go around instead of "waiting" for the car in front.
is written after Z
In general, although there are still many areas for improvement in Xpeng City NGP, its shortcomings are still not concealed. The performance of Xpeng City NGP is considered top-notch among current driving assistance systems, and it basically achieves autonomous driving performance. Furthermore, He Xiaopeng's willingness to expose weaknesses has also become the driving force for Xiaopeng's NGP performance to continue to improve.
To be honest, Da Shijun used to be a consumer who liked driving but did not value driving assistance systems. However, after seeing the practicality of Xpeng NGP driving assistance systems, his heart was indeed shaken. Da Shijun believes that with the further improvement of Xpeng NGP functions, autonomous driving will not be too far away.
(A small program has been added here, please go to the car client to view)
