Autonomous driving – the future of the car track

If the engine is the heart of the fuel car, then the self-driving technology is its soul.

Compared with fuel vehicles, the essential change of new energy vehicles is the change of energy utilization mode, and intelligent technologies such as automatic driving are the most subversive.

In the past few years, the autonomous driving industry has experienced ups and downs, after experiencing a series of turbulences, it has entered a new stage, the rise of Zhixing in the field of unmanned distribution, Xiaoma Zhixing tried heavy trucks, Wenyuan Zhixing tried minibuses, Yuanrong Qixing, light truck business, and gradually changed the pattern of the track.

Discover the "watershed"

After decades of development, driverless cars have come close to practicality, but they still do not reach the level of autonomous and safe driving cars. At present, all players are trying to break through the technical bottleneck and become the first person in L5, but now it seems that the situation is not optimistic.

Waymoone is the world's first driverless commercial service project, focusing on the field of driverless for ten years, is a milestone in the entire field of unmanned driving, but Waymo started in Phoenix and is currently only landing in one corner of San Francisco. Musk has said that Tesla will launch an L5-class Robotaxi in 2020, which will produce 1 million vehicles a year, while Tesla has not built a single car in the past two years. Drive.ai, Baidu's chief scientist, Ng Enda, has sold himself to Apple.

The broad business prospects have attracted many investment institutions to "pay" for autonomous driving, and the valuations of related companies have also risen. The dividends of driverless driving are fading, or the bubble of driverless driving is bursting.

Waymo's market capitalization has slipped all the way from $175 billion three years ago to $30 billion, a drop of more than 800 percent. The amount of financing for domestic companies that once emerged in the field of autonomous driving is also gradually decreasing. According to the data of Tianyancha Professional Edition APP, the number of driverless enterprise registrations reached a peak in 2018, and then there was a downward trend.

In this special era, what does technology mean for autonomous driving companies? Is the whole track going to get stuck in a driverless system? From the perspective of the Internet industry, the current automotive industry has fallen into a technology trap, and it is impossible to save automatic driving by relying on technology alone.

Buffett has never invested in tech stocks, so why would he invest in Apple stock? In a roundtable conversation from FanTalk, they mentioned:

The reason why Buffett does not invest in technology stocks is because the "moat" of most technology companies is the ability of technological innovation, but Buffett believes that this is not stable.

According to IDC's analysis, the global compound annual growth rate of autonomous vehicles will reach 18.3% between 2020 and 2024, and the shipment of L1-L5 self-driving cars will be close to 55 million units by 2024, of which L1 and L2 autonomous vehicles account for 64.4% and 34%, respectively.

That is to say, as far as the current development of the industry is concerned, more than 98% of the future self-driving cars are L1 and L2 levels, and L5 is still a fantasy. From the perspective of commercial landing, L-few is not important, what is important is how to achieve efficient landing in specific scenarios.

You know, traffic is not absolutely safe, even if it is an autonomous driving with scientific and technological genes, it is impossible to have a traffic accident 100%, and it will only be infinitely close to "zero", that is, L5. On the one hand, the technical bottleneck is difficult to break through, on the other hand, unlike Apple's product logic, unmanned driving is constrained by safety issues, and it is more difficult to complete the landing of science and technology.

If we have to set a watershed for unmanned driving, the standard of "people" may be the most reasonable.

Taking "people" as the "reference answer", as long as the automatic driving technology can reach or even exceed the safety level of artificial driving and reduce the accident rate, then automatic driving is beneficial to the whole society.

Autonomous Driving: A "Liberal Arts" Exam

In fact, autonomous driving not only faces technical difficulties, but also has social opinion environmental problems in the corners ignored by many enterprises and capital. Both of these problems are important, and solving this problem can even give a certain advantage to the enterprise.

In terms of public opinion, the historical lessons of the online ride-hailing industry are still vivid and vivid.

The Information Research Institute of the Ministry of Emergency Management once released the "Research Report on the Safety Development of China's Online Ride-hailing", saying that the safety of major domestic ride-hailing companies is higher than that of traditional taxis, and the accident mortality rate is 26% lower than that of cruise taxis. But every time there is an accident in the online car, it will be pushed to the cusp of the storm, and even some anchors will pretend to be Didi drivers to attract traffic.

Due to the early neglect of public opinion, Didi paid a heavy price for this, in 2019, Didi invested more than 2 billion yuan in safety, and the safety team at all levels has exceeded 2500 people, that is, one in every five employees of Didi is responsible for safety.

Nowadays, with the popularity of L1 and L2 autonomous driving technology, traffic accidents occur in an endless stream, but the entire automatic driving industry seems to ignore this, or even condone it.

On the one hand, the current automotive automatic driving system is not perfect, can only play an auxiliary role, but due to the enterprise transition publicity, resulting in ordinary car owners and users of insufficient understanding of automotive technology, it is easy to ignore the hidden dangers of the technology itself, resulting in frequent safety accidents.

On the other hand, after the accident, the major car companies were arrogant in their attitude towards the rights protection of the owners, and even did not seriously analyze the causes after the accident, resulting in accidents. For example, after the accident of Weilai car owners, Weilai issued a "joint statement on the cognition of NIO owners on the NP/NOP system" to try to clear up the relationship, and the result was that Weilai car owners spoke out one after another, believing that they were "represented".

The shadow left by unmanned driving in the minds of users will continue to exist, and if it is allowed to develop, it will continue to develop, even if it has matured, it will cost more to solve it. You know, autonomous driving is not absolutely safe, and there are also accidents that may occur on L5, and the problem will be more serious.

From a problem-solving perspective, driverless driving is essentially a "liberal arts" exam, not only technology can not give absolute answers, but the industry is also facing social and legal challenges.

Progressive route

If divided according to the business route of autonomous driving enterprises, it can be roughly divided into two types: one is an L4/L5 level enterprise that ascends to the sky and takes a completely driverless route, such as Waymo; the other is a gradually developing ADAS, such as Tesla, such as Wei Xiaoli, such as the just-rising Zhixing and intelligent driving technology.

Waymo, as the benchmark of the unmanned driving industry, has now shown the bottom card of the pressure box, the technology is difficult to improve, the core talent is lost, the commercialization is difficult, the capital is disappointed, and the followers are gradually differentiated and change their course. The other kind of progressive player has become a fragrant feast in the eyes of capital. For example, before the announcement of this year's financing news, almost not many people knew the name, and as a result, the first round of financing was invested by Shougang's Meituan Hillhouse for hundreds of millions.

How can a player like Mi mo Zhixing rise, and what is the difference between wei Xiaoli and Tesla? Next, he needs to sort out this step-by-step route.

From an Internet perspective, the progressive route is essentially an MVP (Most Viarable Product) design.

With the thinking of the Internet, if you feel that this solution has market value, you can invest some resources to develop a "feasible minimization" product. Prototype in the most concise way possible, then refine the details quickly and respond quickly to user needs by iterating. The so-called "minimization" is to solve the customer's pain points. Its advantages are to speed up the time to market, reduce resource investment, reduce the cost of trial and error, and the disadvantages are poor stability and reliability.

Unmanned driving is a very complicated thing, and everyone only cares about the supply of technology, thinking that as long as technology is not a problem, autonomous driving can succeed. It turns out that compared with the supply of technology, the demand for technology is the real factor that determines the survival of enterprises.

One is "supply", the other is fatal "demand", MVP design is essentially demand, but the MVP design of autonomous driving has diverged.

Some insiders said that at present, most people in the industry think that it does not matter whether there is a bug or not, as long as the car can run, and then repair it if there is a problem. Although Tesla and Wei Xiaoli have not yet fully mastered the self-driving technology, they still apply the self-driving system to the car.

In the field of the Internet, when the MVP was designed, Tesla also encountered similar problems, but the difference is that the mobile phone can be restarted if it is broken.

Unlike Tesla and Wei Xiaoli, the players of the Ultimate Wisdom Walk. Tesla and Wei Xiaoli's products are both MVP-level products. Players like Zhixing are designed to be MVP-level.

Autonomous driving is not a simple technology, but a technology that combines artificial intelligence, visual computing, radar and other technologies.

And compared with progressive products, the progressive scene is more conducive to avoiding the occurrence of safety accidents. For example, for low-speed logistics and distribution vehicles, "low speed" means that even if there is a failure, the loss caused will not be too large. At the same time, "logistics distribution" belongs to the business model, the C-end is not the main user, on the contrary, with the change of the operating model, people's distrust and hostility to autonomous driving technology will also disappear, thus gaining social recognition.

Scene progressive also has a disadvantage, that is, it is difficult to solve the corner event of long-tail data. Security and data are like fish and bear paws, you can't have both.

Sensor solutions and computing platforms are getting closer and closer, and there are fewer and fewer vendors. Therefore, the data determines the victory or defeat of the self-driving company. Long-tail data problems in different scenarios and the degree of difficulty in solving long-tail data problems.

If you choose a scene that is easy to land as a hotbed, it is difficult to obtain long-tail data, and the enterprise is likely to be surpassed; if all the scenes are invested in complex scenarios, it is likely to be unable to make ends meet. The best way is to use simple scenes to "raise a family" and "attack" complex scenes at the same time.

Driverless has reached the time for reflection, the transformation of technology is not far away, and I hope that players can realize their dreams in the future.