Autonomous driving ushered in the era of alliance

Author | Wei Qiyang

Source | Insight New Research Society

Autonomous driving winds are often there, and the wind direction is unpredictable.

Five years ago (2018), a large number of Chinese autonomous driving companies were established at this time node. At that time, the most important thing in the industry was to cooperate with the OEMs to make DEMO samples show muscles. Even Baidu cooperated with Jinlong Bus to produce. A driverless microcirculation electric vehicle Apolong, Li Yanhong positioned it as "the world's first L4 mass production autonomous driving bus".

Two years ago (2020), after Baidu opened Robotaxi on Changsha (2019) on the road, Wenyuan Zhixing, Xiaoma Zhixing, Autox, Yuan Rongqixing and other autonomous driving companies opened the large and small Robotaxi teams into the team. For all cities across the country, for a while, Robotaxi became the absolute top flow of the autonomous driving industry.

One year ago (2021), Tusson successfully landed in U.S. stocks in the future, causing the industry to be very enthusiastic about autonomous driving commercial vehicles. Baidu ecological company Deepway was hot while hot, and the intelligent heavy truck was released -Star Road 1st, Pony Zhixing released a truck business The independent brand "Pony Smart Card", followed by the executive team to run away, and the three companies, three companies, cracking thousands of hanging technology, Xingxing Oranges, and Optimus Smart Card, all focused on autonomous truck tracks.

Unlike Robotaxi, Robotruck has also attracted the participation of the OEM. By establishing the form of a autonomous driver's subsidiary, the relevant business is promoted. Tuyou is the representative of them.

Now (2022), players in the industry have played a game of "reduction of dimensions" and "convergence", from L4's self -reduced body to do L2, Wen Yuanzhi, who originally did Robotaxi, began to cooperate with Bosch to do it as a thing as Bosch. The car company delivered L2+auxiliary driving, and the light boat Zhizhou Zhihang, who originally made Robobus, released a low -cost L2 mass -produced kit ... In particular, Baidu and Huawei entered the car in different postures, which stirred a larger wave in the industry.

In the process, the autonomous driving company, the OEM, Tire1, and the scene party formed a firm interest alliance, and the new pattern of autonomous driving gradually took shape.

01 Chasing the wind and freedom

When it comes to autonomous driving, it is easy for us to be linked to redefining cars and changing travel methods such as temptation and imagination. However, the initial cause of human development of autonomous driving is because of war.

In 2001, the United States was trapped in the mud pond in Afghanistan. In order to deal with a large number of casualties caused by roadside bombs, the US Congress requested that in 2015, one -third of the military's vehicles must be driving.

The change is always faster than the plan. In 2003, the United States was involved in the Iraq war, but the progress of unmanned driving was very small. Under the chattering questioning of the Congress, DARPA (DEFENSE Advanced Research Projects Agency, the US Department of Defense Advanced Research Planning Planning Bureau ) As soon as the clever machine moves, it is used to promote the research and development of technology in the form of unmanned driving competition. As long as the participating team can cross the Mohavi Desert from Los Angeles to Las Vegas, you can receive a $ 1 million bonus.

A total of three sessions were held in the competition.

In the first session, all the teams returned tragically. No one completed the game. The Kanegielon team who walked the farthest walked only 10 kilometers, less than 5%of all schedules.

In the second session, someone finally finished the game. The Stanford team became the champion of the competition because of the addition of visual solutions.

In the third session, the game was difficult to upgrade and the bonuses doubled. In addition to the desert section, the city section was also added. The final results showed that the results of the participating team have greatly improved compared to the previous session.

In these few competitions, the two founders of Google, Sergei Brin and Larry Page, also ignited their strong interest in driving unmanned driving. After a series of talent recruitment, Google’s The autonomous driving project is officially launched in 2009, which is now the predecessor of Waymo, a well -known Waymo.

In 2016, Waymo spin -off from Google. Soon after (2018), Morgan Stanley gave $ 175 billion in sky -high valuations. This is the valuation of Uber, Airbnb, and WeWork, which was born in the financial crisis in 2008 in 2008. There are a lot of up to the market value of Toyota Motor, the world's largest car manufacturer.

At this time, the autonomous driving was also wearing a veil, which was both mysterious and noble.

In addition to capital, there are a large number of top -level technical talents. Around 2018, a large number of autonomous driving companies were established in China. They checked the background of the main founders of these companies. The first batch of self -driving/artificial intelligence core technical talents almost supported half of the Chinese autonomous driving industry.

Source: Network Comprehensive Table: Insight, New Research Society

Most of these top -level technical talents have the labels such as "technical believers". On the surface, they are engaged in autonomous driving to realize "personal interests and ideal ambitions". The self -confidence of who I am, instead of spending passion in the comfort zone, it is better to start a world of entrepreneurship.

In their eyes, the absolute right to speak is absolute control and freedom of autonomous driving technology. However, there are always differences between ideals and reality, and the commercialization process of autonomous driving has not been performed as expected.

In 2017, Waymo began to land in the Robotaxi business in Phoenix (at that time there was a security officer): to expand to 9 cities in 18 months, Waymo said, "It has solved the problem of 99%of driverless driving 99% ","

But 5 years have passed, and the 1%problem is still not. Waymo's 18 -month plan that year did not see the hope of realization.

In 2018, Waymo stated that he had signed a contract with Jaguar and will produce 20,000 Jaguar SUV autonomous cars; a few months later, Waymo will release the news that it will cooperate with Chrysler to produce more than 60,000 autonomous vans.

Of course, these "blowing cows" have not been realized until now, and the entire industry has begun to reflect. Is it too optimistic about autonomous driving?

After Uber's accident in 2018, the outside world had a clearer understanding of autonomous driving. Chinese autonomous driving companies also became cautious, no longer "painting cakes", but sinking down to polish technology and studying the landing route.

02 arrogance and bias

Chinese autonomous driving technical brothers, all of them are genius, and geniuses are often very arrogant. Therefore, in the autonomous driving circle earlier, before the absolute technical authority was established, "dissatisfaction with each other" was The main theme between friends and merchants. In the dictionary of the technical big brother: cooperation is impossible, and it is impossible to cooperate in this life.

For example, in terms of technical routes, Momenta has introduced the concept of "flywheel" L4 in early 20120, that is, the continuous accumulation and iteration of the two closed -loop automation of the two -level autonomous driving is promoted by mass production data, data -driven algorithms, and closed -loop automation.

The controversy lies in Momentaceo's interpretation of data sources:

MOMENTA is except for simulation by carried the MPILOT MPILOT on the vehicle to solve problems and create value for customers, and recycled massive data to better iterate products ... Promote the continuous upgrade of MSD, so that no one is completely unattended at all The driving system is constantly evolving.

Explain that MPILOT is a mass -produced autonomous driving of Momenta. Although it has high -level autonomous driving capabilities, at this stage, it is mainly facing L2 -level autonomous driving. MSD is the complete driverless solution of Momenta.

Momenta's "chicken thief" is to use L2/L3 data to "feed" L4 -level system to promote iteration. Can low -dimensional data be applied to high -dimensional application? Many technologies do not agree.

Yu Yan, CEO of Qingzhou Zhihang, believes: "From the types and computing platforms of sensors, the application scenarios of L2/L3 are very different from L4, and it is difficult to form interoperability with the data of L4. Big challenge. "

Xiaoma Zhixing believes: "It is relatively easy to use a high -profile L4 -level autonomous driving architecture to do the L2/L3 level. Conversely, low configuration data is difficult to use for testing and certification of the entire L4 level system."

Wang Naiyan, the chief scientist of Tucson in the future, directly named Momenta. He wrote a special article and pointed out that "in some manufacturers and some media's intentional or unintentional rendering, the data is the pirate of Ling Dan. A L2 system can "grow 'into a L4 system ..."

However, the actual situation is that a reliable L4 system requires the cooperation of engineering, algorithms, hardware, and vehicles to create.

Then on the route of commercial landing, the views of each family are not consistent.

According to Momenta's "flywheel" L4 concept, its commercial landing is divided into two stages. The previous stage is equivalent to mass -produced products that ADAS assisted driving, and the ultimate pattern Robotaxi of the autonomous driving is only in the latter stage.

In this regard, Han Xu, CEO of Wenyuan Zhixing, put forward the famous "Bunie Rabbit without a rabbit" in front of Cao Xudong. The crushed silver was not picked up, so I cut my thorns, opened the road every mountain, and met the water bridge, and I built the autonomous taxi team along the way. "

Now, in fact, whether it is Momenta or Wenyuan Zhixing, they have corrected their own business routes in the later period. Momenta directly across the L2 stage and formed its own Robotaxi team. Exploration in one direction.

The summary of Baidu's experience is: "Climbing Everest, getting eggs along the way".

In addition, in the same segment track, friends and merchants are even more different.

In the field of Robotruck, because L4 -level autonomous trucks cannot be commercialized in the short term, the company represented by Supu Technology has proposed a compromise plan, that is, the L3 -level autonomous truck is first landed. Double -driving "change" single driving.

The point of view of Qianjian Technology is that L3 -level autonomous driving can greatly reduce the workload of the driver. Originally, a driver had to rotate for 4 hours. In the L3 -level autonomous driving session, a driver may be allowed to drive 8 hours before get off work. Chen Mo, the co -founder of Tusson in the future, uses "nonsense" to evaluate this model. In an interview with "Late LatePost", Chen Mo described the scene of the truck driving in detail: why sports cars are generally husband and wife, father and son, or brothers. Two drivers rotated because one person was always resting and sleeping.

"All drivers are rotating in 4 hours. The two drivers are rotating and sleeping. This is how the sports car runs ... This car usually runs for 14 to 15 hours a day. I, can this driver open for 16 hours because you pretend to be your system? Are you legal? Can you hold you for 16 hours? "

03 Alliance solve anxiety

Although the dispute is still continuing, due to the continuous decrease in the expectations of autonomous driving commercial landing, autonomous driving companies that have previously believed that "single -drying" can also start to reflect on, adjust, or correct their own technology and business routes. A clear wind direction is that there are more and more dimensions of autonomous driving companies and more and more cross -border alliances.

In the past, the L2 and L4 routes did not offend the river water. The two were not only clear, but also despised each other.

L2 believes that L4 is unrealistic and not grounded; L4 believes that L2 is a low -dimensional application and crushed silver on the road.

But the border between the two is being broken today, and the main factor that promotes the cross -border of the autonomous driving company is "alliance".

In May of this year, Wen Yuan Zhiyi, who firmly took the Waymo route, got investment in Bosch, and the two parties began to cooperate to deliver L2+auxiliary driving to the car company; L2 mass -produced kit, its main partner is Dongfeng Yuexiang and the horizontal line of the autonomous driving chip company; Momenta has cooperation with BYD and SAIC ...

Why does the attitude of autonomous driving companies come to "big turns"? This has to start with their situation.

Let's take a look at the external environment first.

Shortly after the listing of Tusson in the future, it was initiated by the US Foreign Investment Commission (CFIUS). Tucson had to transfer some technical supervision power of the autonomous truck business to the US government in the future. At the same time, it adopted a technical control plan. Limit the access to the Chinese branch to the autonomous driving data. The R & D and business of China and the United States gradually stripped independence.

Another case, due to changes in the regulatory environment of the United States, the SPAC listing plan of Zhiga Technology was forced to suspend in November last year.

For Chinese autonomous driving companies, it is not possible to go public in the United States. In addition, the valuation of the head autonomous driving company represented by Xiaoma Zhixing and Wen Yuanzhi's behavior has been in a very high position. After the first delivery of round D financing, the valuation was 8.5 billion US dollars. After the completion of the C -round C financing in May this year, Wenyuan Zhixing and Fortune exceeded 3.3 billion US dollars.

The financing of other autonomous driving companies is basically above the C round, and most of the valuations also exceed 10 billion yuan.

This also means that if there is no better story and cannot be listed in the near future, it is difficult to have capital to open a new round of financing. When the passage of blood transfusion is blocked, the autonomous driving company must consider the way to survive.

At this time, driven by the new forces of Tesla and car building, L2 auxiliary driving ushered in high -speed development. Among the multiple realization paths, mass production cooperation with the OEM is undoubtedly the fastest. of.

In the end, even if the autonomous driving company is still carrying the beautiful goals of L4, it will eventually enter the river of "car building". Without the support of the OEM, it will never be successful.

In fact, American autonomous driving companies have made a good demonstration in the cooperation between commercial landing and alliances.

Three American car companies, GM, Ford, and Chrysler, formed a pattern of Waymo Union Chrysler, GM Union CRUISE, and Ford Union AI AI unmanned alliance.

Both car companies and autonomous driving companies realize that unmanned driving is not a "solitary". The two parties must complement the resources, technology, talent, management and other aspects. Technology companies have rapidly iterative management methods; autonomous driving companies need to understand the inherent problems of the automotive industry and be able to integrate the industrial chain.

The autonomous driving company takes the initiative to show a step of "alliance" from the anxiety of its own survival, and the OEM, Tier1 and Scenario are willing to form an alliance with the autonomous driving company to ease anxiety.

In addition to complementary advantages with the autonomous driving company, the OEM has also been promoted by the intelligent change of automobile products. The supply chain ecosystem that supports the development of the industry has also changed. Essence

For example, in the past, the OEMs and Tire1 and Tire2 were one -way linear relationship, and now they have evolved into a more complex new network system;

For example, in the business model, the car hardware was directly sold directly, and now it has gradually evolved into hardware plus subsequent software services;

For another example, the automotive research and development process, from the integration of software and hardware in the past, transformed into software and hardware decoupling and development.

The importance of software is becoming more and more prominent, which has also become a link between the OEMs and autonomous driving companies. Tire1's anxiety lies in the changes in the supply chain system to the challenge to its own position. Instead of passive changes, it is better to actively integrate.

Although Tire1 has the ability to develop and mass -producing production of L2 -level auxiliary driving technology, from a longer -term perspective, if L2 to L4 is opened, some of them are undertaking and reused, Tire1 and autonomous driving companies. Cooperation is still valuable.

As soon as several parties were shot, the era of alliances was produced.

04 The next air outlet

The current pattern seems to be happy, but no one can guarantee whether it will be regenerated in the future. No one knows how far is the current Mount Everest who is unmanned. Therefore, it is very important to maintain competitiveness outside the alliance.

There are two examples here.

One is Huawei, which is "not building a car", and the other is Baidu Apollo, which has a full stack layout.

Huawei's strategy is mainly focused on the moment, helping car companies to build cars in ALL in, and launched a "vehicle solution" around chips, modules, terminals, pipelines, and cloud platforms. Waiting for the cooperation of car companies, Huawei showed strong product capabilities. Among them, the M5 M5 87 days had been delivered to 11,296 units, setting the fastest record for the delivery of new brands of single models.

The "translation" of this case is that no matter how the industry changes, the autonomous driving company must let the "allies" feel the dividends brought by the empowerment of dimension reduction technology and form the core competitiveness of "alliance" with the car company. It goes far enough in the long march of autonomous driving.

Baidu Apollo's strategy is to take into account both the following and the future.

In September 2019, Baidu's ROBOTAXI team opened the streets of Changsha and became the first Robotaxi team to carry manned test on open roads. After Baidu, the Robotaxi team in domestic cities with a 2G mode was set up in a 2G mode. During the time, it became a realization way for autonomous driving companies.

In September 2020, Baidu released the "5G Cloud Delivery" at the "Baidu World 2020 Conference", and immediately set off a trend of inviting security officers in the industry. A deeper impact of technical capabilities is a clearance of the industry bubble.

In June of this year, Baidu Apollo announced that its autonomous driving service platform "Radish Run" opened a pilot pilot of commercialization in Wuhan Economic Development Zone, providing the public with autonomous driving paid travel services. Earlier, radish running has been realized in the commercialization of Beijing, Chongqing, and Yangquan, and Wuhan is the fourth city.

At the Baidu World Conference not long ago, Baidu Apollo launched the sixth -generation production unmanned car Apollo RT6 with a cost of only 250,000.

On the landing of L4 -level applications, Baidu maintains an absolute lead. After creating a model case every time, it will attract a large number of friends and merchants, while Baidu finds the scene in the process. Copy replication. The cooperation here is not only the car, but also the full stack technology related to autonomous driving.

Like the intelligent transportation field related to the car roads, Baidu Apollo enters the bureau with new infrastructure participants and won 460 million yuan in 2020 "Guangzhou Development Zone, Huangpu District, Guangzhou, Guangzhou Development Zone for autonomous driving and car roads. The project "large order is a sign, and quickly establish an authoritative image, becoming a pivotal in the field of smart transportation.

The reason why Baidu can fully bloom in various areas of autonomous driving is in addition to the foundation of the technical foundation, but also the mutual strength of ecological allies. According to the official data of Baidu Apollo, of the more than 200 ecological partners, there are both mainstream automobile manufacturers, as well as some of them. The well -known first -level parts suppliers, in addition, have chip companies, sensor companies, transportation integrators, travel companies, etc., which are closely related to smart transportation new infrastructure and autonomous driving, covering the complete industrial chain from hardware to software.

Back to the chase of autonomous driving wind direction, there is Baidu's "homework" ahead, and "copy" is finished.

Technically, the high -dimensional research and development of L4 level is the foundation and the foundation of the foothold. No matter how the future wind direction changes, as long as the high -dimensional technology can be competitive, it will naturally not be easily out.

When the application landing, keep the follow -up of Baidu. Sometimes, slow is fast. The trial and error experience of Baidu may be sent first, not to mention, the cake of autonomous driving is large enough. Even if it is only a small piece, the market capacity is starting 100 million.

Finally, remember a principle. With the removal of autonomous driving, its reality and dilemma are gradually cut in front of the public, all technical routes and business logic must be proceeded from pragmatic, and "when can autonomous driving be realized" The question is aside. Before making any decision, ask yourself to benefit?

Reference

1. Late LatePost, fifteen years of unmanned driving in the United States: war, betrayal and Google past

2. The sword of Yigongzi, Waymo, Baidu, autonomous driving kills the "superhero"

3. Silicon Star, 13 years of autonomous driving, only blooming and not results

4. Tiger sniff, genius, flicker and cannon fodder

5. Smart Driver Network, Momenta drone crossing through urban and rural binding departments: mass production sensor solution, four -year bicycle profitability 6. Smart car reference, Xiaoma Zhixing: 3 years of production of Robotaxi!Unmanned truck plan is fully exposed for the first time

7. Wang Naiyan, is massive data a silver bomb for high -level autonomous driving?

8. Yuanchuan Research Institute, self -driving removing charm: From dimensional reduction to diminishing dimension reduction

9. Eu Automobile, Huawei, take the lead in disruption

- END -