Text: Original by Understanding Car Emperor Liu Yiwei [Original by Understanding Car Emperor Industry] Editor's note: 2022 has come to an end. Looking back on this year, the automobile industry is intertwined with black swans and gray rhinos, and oil prices and car prices have th

article: Original by Understand Car Emperor Liu Yiwei

[Original by Understand Car Emperor Industry] Editor's note: 2022 has come to an end. Looking back on this year, the automobile industry is intertwined with black swans and gray rhinos, and oil prices and car prices have their own ups and downs. Standing at the end of 2022, we are about to witness the withdrawal of 14-year-old new energy subsidies, and we will also witness the end of preferential policies such as the reduction of purchase tax as 2022 ends. In order to review the ups and downs of the year 2022, Undercar Dite launched the annual series planning "2022·Car Market Chronicle". We hope to reshape the records of the eve through the automotive pattern and provide an anchor point for people to review history in the future. This is the fourth article in the series: "2022·Car Market Chronicles | Assisted driving systems cost tens of thousands?" Giants in the autonomous driving industry are making headlines one after another》

This year, high-end intelligent driving has gradually begun to be mass-produced, especially the urban NOA function has become the focus of various car companies. Car companies including Xpeng , Jihu, and Weipai have successively announced that they will mass-produce urban pilots. is in sharp contrast to the car companies' vigorous promotion of high-end intelligent driving. The user experience of the intelligent assisted driving functions already on the car is not ideal. Behind the high price, common functions are not used frequently, and higher-level functions basically belong to " futures ".

End users do not pay the bill, and it is more difficult for autonomous driving companies as solution providers to make profits. As a result, the capital market is gradually returning to rationality. Data from the RuiShou analysis platform shows that there were 67 financing incidents in the domestic autonomous driving field from January to October this year, with the total disclosed financing amount being 14.3 billion yuan, a year-on-year decrease of 32% and 61% respectively. Without funding sources, layoffs and salary cuts have become the main theme of the entire industry, and autonomous driving is heading towards the cold winter.

The giants in the autonomous driving industry are making headlines one after another.

Is it expensive and difficult to use? Is it worthless intelligent assisted driving?

What is the current level of autonomous driving capabilities?

From a publicity perspective, many car companies define their solutions as between L2 and L3. To be more precise, L2-level intelligent driving assistance systems are still the mainstream among currently mass-produced and available models. From the perspective of functional configuration, the more common intelligent assisted driving mainly focuses on the two major functions of high-speed navigation assistance and intelligent parking.

For users, having an intelligent assistance system means paying a higher price.

The public selling prices of car companies show that the above-mentioned assisted driving systems in model options can easily cost tens of thousands. For example, the original price of Xpeng's XPILOT 3.0 intelligent driving assistance system was as high as 36,000 yuan (later adjusted from "paid upgrade" to standard), the prices of NIO 's NIO Pilot full package, and Tesla 's EAP are respectively 39,000 yuan and 32,000 yuan.

In addition, there are higher-level functional upgrades, mainly focusing on urban intelligent navigation assistance. However, not many have actually implemented it, and more car companies have chosen to use it as publicity and make money by buying "futures".

The most representative one is "Tesla FSD", which also includes the urban smart cruise of Jifox Alpha S Huawei HI version, the complete package of ZAD intelligent driving assistance system of Jikrypton Automobile, etc. These functions can be purchased in advance and subsequently upgraded through OTA.

Source: Reuters

Taking the latest version of Tesla FSD as an example, the price is now as high as US$15,000 (approximately RMB 104,000), and the available functions are limited, far from achieving full autonomous driving in the true sense.

If the price has deterred many consumers, then put aside the price factor and ignore the "incomplete" configuration. How often do users who have experienced these functions actually use it?

Xpeng G9

Taking Xpeng Motors as an example, during the Spring Festival of 2022, the driving mileage of Xpeng NGP only accounted for 1.6% of the total mileage, while the ideal NOA figure was 3.7%. This means that the intelligent assisted driving functions currently promoted by car companies are not used frequently in real scenarios.

The reason can be referred to a comment from a netizen: "One is because it is troublesome, and the other is because I can't believe it."

The former is not difficult to understand. New functions will be introduced online and users will need to spend time learning them, while the latter is due to the fact that autonomous driving technology is not yet mature.

The scene where the owner of a Xiaopeng car hit and killed someone

In August 2022, Xiaopeng P7, with assisted driving turned on, hit a malfunctioning vehicle parked on the elevated road at a speed of 80 kilometers per hour. A man was knocked away at the scene, causing casualties. Data released by U.S. safety regulators show that there have been about 400 collisions involving vehicles equipped with partially automated driving assistance systems in the past year.

Yu Chunjun, deputy director of the Institute of Traffic Management Science of the Ministry of Public Security, said bluntly that in recent years, there have been many similar vicious traffic accidents involving assisted driving at home and abroad, and there are indeed loopholes in the vehicle perception and decision-making algorithm.

In addition to the above, from a consumer perspective, once you choose to purchase a vehicle with higher-end intelligent assisted driving functions, in addition to accepting visible expenses and unstable "functional experience", you must also be prepared for continued battles, such as lidar , camera and other hardware replacement, as well as high maintenance costs.

After decades of dominance, autonomous driving has entered the cold winter.

The autonomous driving industry has changed. This cold wave has brought down many industry giants.

Picture source: Reuters

html In August, self-driving US stock company Aurora Innovation revealed that it plans to lay off employees and sell some assets to survive; in October, L4 self-driving unicorn Argo, which was once valued at more than 7 billion US dollars, AI officially announced its closure; Waymo, a subsidiary of Google , is one of the giants in the autonomous driving industry. Its market value has shrunk significantly, from a peak of 175 billion US dollars to 30 billion US dollars.

This cold air has also blown into the country from the other side of the ocean. There were reports that China Automotive Intelligent Control, a subsidiary of China Automotive Connect, began to lay off employees, including many in management. Later, Pony.ai was exposed to internal layoffs, and individual departments were directly disbanded. Although domestic autonomous driving companies that started relatively late have not yet gone bankrupt, layoffs and business tightening have occurred from time to time.

Pony.ai Unmanned Vehicles

All signals indicate that the current self-driving companies have to tighten their belts. To be more precise, it is difficult for the capital market’s financiers to pay for the “story” anymore. Data from the

RuiShou analysis platform shows that there were 67 financing events in the domestic autonomous driving field from January to October this year, with the total amount of financing disclosed being 14.3 billion yuan, a year-on-year decrease of 32% and 61% respectively.

Looking at the overall market, as a capitalist with a very keen sense of smell, he is unwilling to continue to spend money on the autonomous driving industry. The main reason is that there is no hope.

It is difficult for autonomous driving companies to make profits

First of all, R&D costs money and it is too difficult to make profits. Although the enthusiasm of the capital market is not as good as before, most autonomous driving companies are not ready to "wean off".

Informed sources revealed that the current biggest revenue for industry giant Waymo is the compensation from the UBER patent lawsuit. At present, the average revenue of the main business of the top ten autonomous driving companies in the domestic industry does not exceed 300 million yuan, which is far more than the amount spent by ordinary autonomous driving companies. These include the labor costs, equipment costs, R&D experiment costs, and the cost of collecting massive data for self-driving algorithm engineers that cost millions.

It is too difficult for self-driving companies to make profits

Self-driving developers said that self-driving requires vehicles as carriers, and manufacturers cost money to produce cars. Whether they use existing vehicles or specialize in small-scale production, the cost is not low. This resulted in many autonomous driving Tier 1 companies burning a lot of money, and ultimately failed to form a profitable business model.

In other words, for the vast majority of autonomous driving companies, financing is still an important source of current company income. This means that the company's power of life and death is closely related to the capital market.

Secondly, it is too difficult to implement L4 autonomous driving technology.

From the perspective of the implementation path of autonomous driving, the industry is mainly divided into two factions. One is a progressive technology route represented by Tesla, which advocates iterating from L2 to L4; the other is a leapfrog technology route represented by Waymo, which will directly implement L4 autonomous driving in the future.

L4 level autonomous driving technology is too difficult to implement.

Both sides have their own opinions. The former is full of scraps of money, while the latter is pursuing a blockbuster. What the two have in common is the hope to achieve L4 autonomous driving.

L4 level of autonomous driving means that it has entered the category of highly automated driving, and the car owner does not need to control the steering wheel. In terms of liability definition, once a traffic accident occurs, L4 Autonomous Driving Company is fully responsible for the accident.

The Ministry of Industry and Information Technology released the national standard "Vehicle Driving Automation Classification"

To promote the commercialization of autonomous driving, data is the highlight. Yu Qiankun, technical director of SAIC Group's Artificial Intelligence Laboratory, told Chedi.com that the machine learning algorithm needs to iterate with a large amount of real data, and real-world scenarios are uncertain, and there will always be corner cases that cannot be covered by the algorithm. At this stage, it is difficult for autonomous driving to surpass the level of human drivers.

Yu Qian, co-founder and CEO of Qingzhou Zhihang, said that the difficulty of driverless driving is that the driver does not need to interfere with the car in any case. This leap is a fundamental leap. Whether the gap between technology and application can be bridged does not depend on a certain company, but on the substantial advancement of the entire industry, including sensor technology, computing technology, algorithm improvements, people's acceptance, and the improvement of laws and regulations.

is written at the end:

Finally, tell a little story.

An employee of Waymo, a subsidiary of Google, revealed on social platforms that Waymo’s current solution can basically complete L4 fully driverless driving, but the factors that restrict promotion are nothing more than background rescue resources and hardware costs, and it is still unable to achieve positive cash flow .

When asked whether the current level can be considered as L4 autonomous driving, the other party said: "Does it matter whether it is L4 or not? The key is that the accounting is not correct, understand?

It is foreseeable that the domestic autonomous driving industry is about to undergo a major reshuffle. For enterprises, before the crisis officially comes, they must not only go on the right path, but also seize the time to launch commercial products that make the OEMs pay for them and consumers trust them. After all, only those who are alive are qualified to talk about the future.