Staying up all night, piling up crowds, car companies are crazy about intelligent driving, how many engineers' hair has been swept away?

Staying up all night, piling up crowds, car companies are crazy about intelligent driving, how many engineers' hair has been swept away?

Staying up all night, piling up crowds, car companies are crazy about intelligent driving, how many engineers' hair has been swept away?

After Zhijie S7 became popular with an unmanned valet parking video, car companies such as Xiaopeng and Ideal also quickly followed suit, showing off their smart driving muscles.

This year, car companies first rolled up the number of smart driving in the city, and then rolled up subdivision technologies, such as no map, AEB, and unmanned valet parking. The speed of technology iteration is getting faster and faster, which is staggering.

However, what is not known is that behind the high-frequency update of technology, there are countless intelligent drivers who stay up all night. Engineers sacrificed sleep and hair to keep up with the project. Most of them have to work overtime until at least 10 p.m. before they can go home, and some people sigh, "There is no one who does not roll up if they are engaged in intelligent driving."

The automotive industry has entered the second half of the race, and intelligent driving has become a battleground. The competition is too fierce, and car companies can only rely on crowd tactics to strive for the speed of technology iteration.

The end of the universe has not yet arrived, and the involution of intelligent driving is endless.

Text | Wang Xiaojuan

Edit | Li Huanhuan

Operations | potato

There is no most volume, only more volume

After leaving work at 1:30 a.m., Li Junyi had to commute for 40 minutes to get home. After scribbling and washing, it was already after 2 o'clock in the night when I climbed into bed. Even if I leave work a little earlier, I will play two decompression games, and I always have to stay up until 1 o'clock before I can go to bed.

Staying up late doesn't mean you can relax the next day. At 9 o'clock in the morning, Li Junyi appeared in front of the computer again to start a new day's work, more often, people have returned to their workstations, but the sleepiness of the previous night has not disappeared.

For Li Junyi, such high-intensity overtime is becoming more and more frequent - overtime until 9 or 10 p.m. is a regular operation, and it is common to go home at 11 or 12 p.m., and now even if he has to work until the early hours of the morning, he is not surprised.

Staying up all night, piling up crowds, car companies are crazy about intelligent driving, how many engineers' hair has been swept away?

▲ High-intensity overtime. Photo / Visual China

Li Junyi works in Huawei's intelligent driving department, where volume has become the norm. Over the years, he has seen hospitalized colleagues called back to work by a phone call, and he has also witnessed the project PL (project team leader) on the second day after kidney stone surgery, he immediately plunged into the high-speed "running" project, "without rest at all, working overtime until after 11 o'clock every day".

Collective involution, everyone's hard work was not in vain. Today, Huawei is the leader in the field of intelligent driving.

At the end of November, a video of the Zhijie S7 (the first model jointly created by Chery and Huawei's smart car) showing the unmanned valet parking function went viral. In the video, the driver gets up and gets out of the car after completing the pre-operation, and the subsequent series of actions are completely completed by the vehicle itself. I saw it looking for a parking space all the way in the basement, and on the way it also took the initiative to avoid pedestrians and oncoming cars, reversing into the garage, two completed, the action was silky and not stuck, which was really amazing. Netizens commented in amazement: "If it had been a year ago, I would have absolutely doubted that someone was operating it remotely." ”

At the launch of the Zhijie S7, Yu Chengdong, Executive Director of Huawei, CEO of Device BG, and Chairman of the Intelligent Vehicle Solution BU, mentioned that HUAWEI ADS 2.0 "can now be iterated every five days." Behind the technology of "far ahead", it is the crazy involution of every intelligent driving worker.

Staying up all night, piling up crowds, car companies are crazy about intelligent driving, how many engineers' hair has been swept away?

▲ Concept art of HUAWEI ADS 2.0. Photo / Zhijie Automobile's official Weibo

But it's not just Huawei.

On social media, netizens jokingly called the intelligent driving industry the "volume king industry". Mr. Lee's friend works for Momenta, a self-driving company. Describing his friend's work, Li Junyi lacks vocabulary, and can only be said to be very volume, and the work under high intensity and heavy pressure even makes people's appearance "directly fast-forward 10 years, born in 94, looks like 84 years, and his friends ridicule him as a 'rich old man'".

A headhunter looking for candidates for autonomous driving-related positions said that the candidates he contacted came into contact with were from different companies, and "very rolled" was the unanimous feeling of everyone, and it was difficult to say which one was more rolled, "I can only say that there is no mainstream manufacturer that does not roll".

Xiaopeng, who has the confidence to wrestle with Huawei, has R&D teams in China and North America, and due to the time difference, the teams on both sides work in "two shifts" in turns, and the domestic team can directly test the software sent by the North American team in the morning, which is perfect and seamless.

Although NIO does not emphasize the ability to drive intelligently, it is just as busy. A person who has Xi in NIO's intelligent driving department said: "My leader is busy like a spinning top. Another intelligent driving engineer who was laid off by NIO not long ago recalled, "I was involved in death, and I worked a lot of overtime on weekends." ”

At the Xpeng Technology Day on October 24, Li Liyun, the person in charge of Xpeng Intelligent Driving, half-jokingly ridiculed that every word of "(He) Xiaopeng (speaking) is my OKR". The industry involution, to a certain extent, is like this, and the goal of frequent updates is issued layer by layer, and finally falls on the working hours of each R&D personnel.

Staying up all night, piling up crowds, car companies are crazy about intelligent driving, how many engineers' hair has been swept away?

▲ Li Liyun talked about intelligent driving at the 1024 Xiaopeng Technology Day. Photo / Xpeng Motors official

Li Junyi believes that although involution may not necessarily bring advanced technology, it must not be able to do not. When a new feature appears in the industry, other car companies often conduct competitive product analysis, and then quickly catch up, so they need a large number of manpower, at least to not lag behind.

An important reason why everyone is crazy about smart driving is that intelligent driving ability is becoming an important factor affecting consumers' car purchase decisions.

This year, after the release of Xpeng G6, the proportion of orders for the intelligent driving version exceeded 70%, and He Xiaopeng, chairman of Xiaopeng Motors, once said that he had seen the inflection point of intelligent driving. After the launch of the new M7, it won more than 100,000 units in less than three months, of which the order rate of the intelligent driving version reached 60%; at the end of November, the new M7 model added two rear-wheel drive intelligent driving versions.

The sales of the new M7 are still jumping upward, and the latest weekly sales data shows that from November 27 to December 3, the sales volume of the M7 was 4,655 units, which has exceeded the 4,047 units of the ideal L7.

Li Xiang, CEO of Li Auto, once said frankly that the former M7 "disabled the ideal ONE", and now the new M7 is menacing again. Ideal has accelerated the pace of development of intelligent driving. At the third-quarter earnings call, Li Xiang said, "I am confident that in the first half of next year, I will be among the first echelon of intelligent driving that has been verified by the market." In order to achieve the goal, Ideal began to expand the intelligent driving team on a large scale, and offered generous salary conditions. Someone jumped from Huawei to the ideal, and the salary directly doubled. On social media, it was revealed that the ideal autonomous driving division has started a "closed development" mode.

When other functions are becoming more and more homogeneous, intelligent driving ability is the main window to highlight the advantages of differentiation, after all, in the limited market, everyone wants to be seen by users.

From the number of Kaesong volumes to the subdivision technology

The volume of intelligent driving continues to heat up.

In the past year, many players have focused on high-end assisted driving on urban roads, and whoever can implement high-end assisted driving in more cities will be able to win more users in regional markets. Therefore, 2023 is also regarded as the first year of the high-level intelligent driving battle of the city.

The first shot fired in the Battle of Kaesong was Xiaopeng. As early as October 2022, Xpeng P5 took the lead in opening urban NGP in Guangzhou.

Staying up all night, piling up crowds, car companies are crazy about intelligent driving, how many engineers' hair has been swept away?

▲ Rendering of Xpeng P5 city NGP. Photo / Xpeng Motors' official Weibo

During this year's Shanghai Auto Show, each company directly rolled up the number of intelligent driving cities. Yu Chengdong announced that Wenjie will launch the "City NCA" function, covering 45 cities by the end of the year, but this number was quickly overturned, and by September, the new M7 with HUAWEI ADS 2.0 "Intelligent driving has been able to be driven nationwide".

As radical as Huawei, there are ideals. Li Xiang set a target in April to open urban NOA functions in 50 cities by the end of the year, and by the Chengdu Auto Show in August, the ideal city opening target was raised to 100 cities, but this is in parentheses - some roads, commuting mode.

Xpeng keeps its mid-year target unchanged – to open urban NGPs in 50 cities across the country by the end of 2023. According to Ren Shaoqing, vice president of intelligent driving R&D at NIO, NIO started the delivery of enhanced pilot-assisted NOP+ in urban areas based on route opening, route sharing, and convergence into a network. For example, in the fourth quarter of 2023, a total of 60,000 kilometers of urban pilot routes will be opened.

The new forces are fighting with the gods, and the traditional car companies and the second generation are not willing to lag behind. This year, brands such as Zeekr, BYD, and VOYAH have begun to make efforts in intelligent driving.

In April this year, at the BYD Group's financial report exchange meeting, Wang Chuanfu, chairman of BYD, was still a little disdainful, "Unmanned driving is nonsense, it is used to fool the people." But soon, Wang Chuanfu changed his mind. Not long ago, BYD's intelligent driving department changed its commander and recruited troops to expand the team. Zhao Changjiang, General Manager of DENZA Automobile Sales Division, said this year: "Our planning institute has a separate intelligent driving team with thousands of people. BYD's latest intelligent driving achievements will be applied to the Denza N7. ”

Staying up all night, piling up crowds, car companies are crazy about intelligent driving, how many engineers' hair has been swept away?

▲ Rendering of DENZA N7 high-speed NOA. Photo / Denza Auto's official Weibo

At present, the battle of Kaesong has been swept up by Huawei to the end of "the whole country can be opened". As a result, car companies are competing in the field of intelligent driving, which has been extended to more subdivided dimensions. In the middle of the year, Xpeng took the lead in proposing a mapless (non-reliance on high-definition map) city NGP, and after that, many players also challenged the ability to do without a map.

Secretly compare technology, compete for the right to speak in the open, and in the intelligent driving track, the scolding war seems to have not stopped. In an interview with the media, He Xiaopeng commented that the "whole country can be opened" by friendly businessmen, "although it is not difficult to drive on the national road, but it is difficult to drive well".

It was also in this dialogue that some of the more secretive technologies were brought to the fore in the midst of the war. At the press conference of Zhijie S7, Yu Chengdong said that Huawei's AEB technology can achieve automatic braking at a speed of 120km/h when facing stationary vehicles. In response to this technology, He Xiaopeng bluntly said: "Friends talked about AEB, and I think 99% of it is fake. Seeing this sentence, Yu Chengdong couldn't sit still, and mocked in the circle of friends, "Even what AEB is, there are car companies who don't understand it at all!" After a few rounds, it was difficult to distinguish the winner from the loser, and finally the two sides shook hands and made peace, He Xiaopeng expressed his gratitude to Yu Chengdong on Weibo, "Sometimes it's easier to become good friends after misunderstandings."

The choking ends, but the volume doesn't end.

Zhijie S7 is the first to be launched before it is launched, all based on an unmanned valet parking video. At the press conference, a Zhijie S7 drove itself on the stage and stopped at a fixed point under the condition of unmanned driving, igniting a new round of war. A few days later, Liu Yilin, senior director of Xpeng autonomous driving products, released a demonstration video of Xpeng X9's unmanned valet parking, which can achieve unmanned valet parking and remote greeting and pick-up in narrow roads and right-angle bends in the parking lot.

Everyone makes fancy moves on the smart driving track, but in fact, there are still restrictions in terms of regulations. An industry insider said that at present, autonomous driving is rolling to the subdivision because the higher-level autonomous driving regulations have not yet been passed, and the subdivision is constantly escalating, which is the "compensation" of higher-level intelligent driving.

Fortunately, there have been some new developments at the policy level. In mid-November, the Ministry of Industry and Information Technology and other four departments jointly issued the "Notice of the Four Ministries and Commissions on Carrying out the Pilot Work of Intelligent Networked Vehicle Access and Road Access". The notice clarifies that intelligent networked vehicle products that are ready for mass production and equipped with autonomous driving functions (i.e., L3/L4 autonomous driving) will be allowed to enter the road in a limited area.

Suppliers also rolled up

Car companies are rolling up in the front desk, and suppliers are also rolling up in the rear.

An autonomous driving practitioner who jumped from an OEM to a Tier 1 complained, "I thought that the old club was enough, but I didn't expect the whole industry chain to be very volatile." ”

Huali works in a company engaged in autonomous driving software research and development and hardware systems in Shenzhen, and has been in the industry for less than three years, and she clearly feels that the industry has become "faster". "Although it has not yet reached the white heat", it is indeed accelerating as car companies are crazy about autonomous driving technology. The upstream suppliers who supply Huali often work overtime until the early hours of the morning.

Based on the optimistic prospect of the autonomous driving track, many parts suppliers are not satisfied with only providing hardware, as far as Huali knows, several are already expanding to autonomous driving solutions. For example, Bosch, Nvidia, etc., have gone down the field to develop intelligent driving software. At the 2023 China Automotive Supply Chain Summit, Yu Xiao, General Manager of Huawei's Intelligent Vehicle Solution BU Intelligent Vehicle Control Domain, said, "It's not enough now, all the functions of the car are still being designed by OEMs. In the future, all third-party vendors can design the software, and that's where the volume begins. ”

At the same time as volume technology, in the supply chain, everyone is also rolling prices.

Ten years ago, the unit price of Velodyne's automotive lidar was as high as $100,000, but in 2020, the average price in the industry dropped to about 10,000 yuan, and this year, the average price of lidar has reached the level of 1,000 yuan.

Staying up all night, piling up crowds, car companies are crazy about intelligent driving, how many engineers' hair has been swept away?

▲ Automotive LiDAR. Photo / Visual China

The decline in prices is the result of the coupling of technological development and market expansion. In the early days, the technical difficulty and high cost of laser equipment were high, and there were few commercial scenarios and were not large-scale, resulting in the price of a single lidar remaining high. With the implementation of lidar in autonomous driving scenarios, the industry has ushered in explosive growth, with large companies constantly joining, rapid iteration of technologies and products, and large-scale commercial applications, so that the cost can decline rapidly.

Thanks to the price diving of key components, when applied to terminals, users can buy vehicles equipped with lidar at a price of 200,000 yuan + or even lower.

In August this year, Xpeng and Didi announced a cooperation between the two sides to launch about 150,000 high-end intelligent driving models. This directly overturned He Xiaopeng's view in 2021 that "150,000 yuan is a watershed for smart electric vehicles".

When the inflection point of intelligent driving comes, the supply chain also takes off.

CITIC Securities expects that, on the one hand, car companies that are deeply engaged in high-end intelligent driving functions and full-stack self-development and use this as a differentiated selling point are the owners and distributors of the new cake, and on the other hand, the supply chain of autonomous driving chips, domain control, sensors, wire-controlled chassis, and automotive Ethernet chips is expected to benefit from the increase in the penetration rate of high-end intelligent driving as a whole.

However, although the cake is large, it is not easy to grab it. Taking LiDAR as an example, the top 3 companies Hesai, Tudatong and Suteng Juchuang together occupy more than 85% of the industry's market share. Tianyancha data shows that there are currently more than 12,000 autonomous driving-related companies in China. Of these 12,000 runners, there are probably very few who can really run.

(At the request of the interviewee, Li Junyi and Hua Li are pseudonyms in the article)

The article is original by each Auto, and infringement must be investigated.