Autonomous driving, a protracted battle among giants

 Although there are many problems to be solved in autonomous driving, this does not seem to prevent the wind of capital from getting stronger.

On October 26th and 27th, the auto intelligence company Ekatong was exposed to raise 1.3 billion yuan, and the Chinese autonomous driving startup Xiaoma Zhixing raised more than 300 million US dollars (about 2 billion yuan).

The commercialization of autonomous driving is also accelerating. On October 21, local time in the United States, Tesla CEO Elon Musk announced that Tesla's Full Self-Driving (FSD) beta software has been pushed to a small number of customers.

Autonomous driving, a function with its own sci-fi attributes, has overcome many problems from theory to technology, research and development to implementation, and policy levels under the joint promotion of the government, automobile circle, technology circle and investment circle, and is stepping into reality step by step.

However, due to factors such as technology, cost and policy, the development of autonomous driving is still a protracted battle.

1. True or false "fully automatic"?

At the moment when the auto industry is moving towards intelligence, autonomous driving is speeding up. Among them, Tesla, a leading player in new energy vehicles, is aggressively promoting the commercialization of autonomous driving and shaping its image as a pioneer.

As early as July 9, Elon Musk said at the 2020 World Artificial Intelligence Conference that he was confident that he would complete the development of the basic functions of L5 level autonomous driving this year. “ Using Tesla’s current hardware, we only need to improve the software. It can realize L5 level autonomous driving ."

According to the classification method developed by SAE International (Society of Automotive Engineers), autonomous driving is divided into 6 levels, L0-L5, and L5 is the highest level. It belongs to unmanned driving and no longer requires human driver intervention. However, most of the mass-produced vehicles on the market are equipped with L2-level autonomous driving, and the commercialization of the L3 level is not yet mature, and the L4 level is mostly in the testing stage, and no car company has achieved mass production in the true sense.

Regarding Musk’s “ epoch-making ” remarks of “achieving L5 level within the year ”, Wang Tianxing (pseudonym), an autonomous driving engineer working for a large domestic automobile group, said bluntly about “deep insight”. Just listen to Musk’s words , not too much. Seriously, "Tesla's water blowing ability and technical ability are equally top-notch. How to define the basic functions? With Tesla's current technical solutions, L5 cannot achieve mass production of passenger cars."

"L2 is human-oriented, L4 and L5 fully autonomous driving do not require intervention, L3 people and vehicles have basically equal rights, and accident responsibilities cannot be clearly divided. Tesla's products have surpassed all L2 products of the same type on the market, but they do not meet the L3 definition and are close to L3 "Wang Tianxing further said.

But at that time, Tesla still quickly set off a capital climax. On the day after Musk proposed to achieve L5 autonomous driving within the year (July 10), Tesla's stock price rose by more than 10%.

An old Tesla owner told "Deep Diving Insight" that he believes that Tesla can achieve L5 level autonomous driving and intends to purchase Tesla fully autonomous vehicles in the future.

On October 21st, US local time, Tesla’s self-driving story was updated: Musk announced that Tesla’s self-driving beta software will be pushed to a small number of customers.

Musk can't wait to start the commercialization of fully automated driving-he said on social media that the price of fully automated driving will increase by $2,000. After the price increase is completed, the price will be increased from the original 8,000 US dollars to 10,000 US dollars.

Has Tesla's self-driving technology really advanced by leaps and bounds?

Judging from the frequent suspicions from the outside world, maybe not.

The National Highway Traffic Safety Administration (NHTSA) stated in a statement that it has listened to a brief introduction of Tesla's fully automated driving system and believes that the system is an extension of driver assistance software that requires manual supervision. In addition, NHTSA unceremoniously pointed out that “the vehicles currently available for purchase cannot drive themselves.”

This also means that Tesla's so-called "fully autonomous driving" may not be the same concept as the commonly referred to as "fully autonomous driving (ie L5 level)".

Some people point out that, according to the standards of the Society of Automotive Engineers, Tesla's "fully autonomous driving" is only considered an L2 "partially automated" system. NHTSA characterizes it as "Autosteer on city streets" and uses Tesla's brand of lane keeping assist functions.

The capital market also failed to give Tesla a particularly high feedback. With the positive stimulus of five consecutive quarters of profit + the launch of fully automated driving software, Tesla's stock price did not rise much, and the response was cold.

"L4 and L5 are still in the exploratory stage, and the upstream and downstream of the industrial chain are extremely immature. I personally believe that the full industrialization of L4 and L5 for mass users will not be realized in 5-10 years, but the life cycle is longer and the imagination space is greater. There are more ways to play," Wang Tianxing said.

2. Pit game

In the face of the enticing “cake” of autonomous driving, both technology companies and car companies have shown unprecedented enthusiasm for the development and commercialization of autonomous driving technology.

Not only Tesla’s bets, but international giants also have frequent layouts: Earlier, Amazon spent more than $1 billion to acquire autonomous driving company Zoox; Google Waymo became a partner of Volvo’s L4-level autonomous driving technology, and Join forces with Daimler Trucks to build an L4 autonomous truck...

In China, in February this year, 11 ministries and commissions including the National Development and Reform Commission jointly issued the "Smart Car Innovation and Development Strategy", proposing that "by 2025, China's standard smart car technology innovation, industrial ecology, infrastructure, regulations and standards, product supervision and The network security system is basically formed. Intelligent vehicles with conditional autonomous driving can be produced on a large scale, and intelligent vehicles with highly autonomous driving can be marketed in specific environments."

Tech giants such as Alibaba , Baidu , Didi and Huawei, many traditional car companies such as Changan, Geely, SAIC and BAIC, new car-making forces such as Weilai, Xiaopeng, Ideal and Weimar, as well as Xiaoma Zhixing and Zhixing. As the main players on the autonomous driving track, the Tripartite Autonomous Driving Company is constantly upgrading and improving the autonomous driving hardware and software systems on the test road, successively launching L2 level mass production models, planning L3 and L4 level mass production models, and launching L4 level autonomous driving test.

Major domestic autopilot players are still placing bets.

Weilai, which is always used to compare with Tesla, introduced Ren Shaoqing, the former R&D director of Momenta, as the assistant vice president and head of the autonomous driving R&D business in August this year. The company also plans to raise US$1.7 billion, one of which is the research and development of autonomous driving technology. According to media reports, Weilai is planning to independently develop an autonomous driving computing chip, promoted by Li Bin, the chairman and CEO of Weilai.

On June 27 this year, Didi’s four-year self-driving business began to open to the public for test rides, and the test was L4 autonomous driving.

In addition, Baidu announced on October 28 that its autonomous driving safety test mileage has reached 6 million kilometers, with more than 100,000 manned tests, and it has been able to achieve safe operation of closed highways on some urban roads and scenic spots in the park.

The brand-new model that Weimar Automobile is expected to launch early next year will be equipped with the AVP autonomous parking technology jointly developed by Weimar and Baidu, which can realize L4 level unmanned driving in specific scenarios.

"The autonomous driving track is the next new energy vehicle, and it will be on fire next year at the latest. " Xinding Capital Chairman Zhang Chi said to "Deep Diving Insight."

In September, Beijing Zhixingzhe Technology, a Chinese autonomous driving high-tech company, announced that it had completed a C+ round of investment of hundreds of millions of yuan, and Xinding Capital was among the investors. Prior to this, Zhixingzhe had received investment from well-known institutions such as Baidu, Shunwei Capital and JD.com.

"At present, with the self-research of automakers, there are only a dozen domestic players doing autonomous driving. The technology is very core and difficult. New researchers are basically unable to enter. But this track will not One family dominates, because each has its own demands, there must be several competing with each other." Zhang Chi said.

3. Tyrant games

It is not easy to run on the autonomous driving track-a big obstacle lies in funding.

In fact, from the perspective of financing, the companies currently at the forefront of the track basically show strong financing capabilities, mostly at the level of 100 million yuan.

For example, in May of this year, Didi Autopilot Company made its first external financing, and the SoftBank Vision Fund led the investment of US$500 million, which was the largest single financing in the domestic autopilot field at that time.

On October 26, Geely Holding Group’s strategically invested and independently operated automotive intelligence company Ekatong completed the Series A financing led by Baidu and followed by Hainer Asia Ventures (SIG). The total financing amount was 1.3 billion yuan. After the valuation exceeds 10 billion yuan.

On the 27th, Chinese autonomous driving startup Xiaoma Zhixing was revealed to have completed a new round of financing of over 300 million U.S. dollars. It was led by the Ontario Pension Fund of Canada and participated in the investment of FAW Group . The accumulated financing amount has exceeded 1 billion U.S. dollars. Worth $6 billion.

These financing cases all reflect from the side that micro-financing may no longer be able to support the development of high-level technology, and autonomous driving track funds are increasingly concentrated, which has gradually become a "local tyrant game."

"Autonomous driving has entered the knockout stage. Those who cannot burn money, have no landing scenes, and have average technical content will leave the track one after another. What can survive is to have core technology, landing scenes, a certain amount of money back, and income. Continued." Zhang Chi said.

The commercialization of low-level autonomous driving has already begun. Take NIO as an example. According to its introduction, all models of NIO are equipped with the NIO Pilot system, which realizes L2 level automatic assisted driving. Under normal sales conditions, the penetration rate of NIO Pilot is about 25%.

However, what needs to be faced is that to promote the development of autonomous driving to a high level, multiple breakthroughs and applications of cutting-edge technologies are required. In this process, a steady stream of funds and resources are swallowed. But at the same time, due to the immature business model, self-driving projects are facing profitability problems, and they are basically at the stage of losing money and making calls, and they still need capital blood transfusions to survive.

Robotaxi (self-driving taxi) is considered to be one of the most likely scenarios for the commercialization of autonomous driving. UBS predicts that by 2030, the global autonomous driving taxi market may exceed 2 trillion US dollars.

Foreign companies such as Waymo, Zoox and Cruise, as well as domestic Baidu Apollo, AutoX, Didi, Xiaoma Zhixing and Wenyuan Zhixing are also fiercely competing in this field.

According to Zhong Hua, senior vice president of Wenyuan Zhixing Engineering, there are five stages in the development of autonomous driving companies, and the fourth stage of autonomous taxis is the fully unmanned test and trial operation in a limited area and time. When entering the fifth stage (to achieve large-scale and large-scale fully driverless operations), profits can be achieved.

However, the high cost limits the commercialization of autonomous driving.

“Currently, L4 passenger cars cannot face individual users, but can only face corporate users. The application scenarios are limited to travel service platforms, trunk logistics companies, etc..” In Wang Tianxing’s view, Robotaxi L4 companies cannot make profits in the short term. To be able to burn money to live, "The main reason is that the cost of lidar can't come down. A car with tens of millions of yuan will hardly be bought by anyone."

Zhang Chi also pointed out that L4 and L5 level automatic driving means that vehicles have to install a large number of lidar sensors, high-precision millimeter wave radars, monocular and binocular cameras, etc., but now the core components, especially lidar, are still difficult to achieve. low cost.

"At present, the cost of transforming a car into L4 or above autonomous driving is about 1 million yuan. The cost is huge, which also means that an L4 and L5 autonomous driving car has to be sold for 1.3 million to 1.5 million yuan to make a profit. The popularity of China will be a huge challenge." Zhang Chi gave an example.

4. Protracted war

At present, the industry has basically reached a consensus: true fully automatic driving is difficult to achieve in the short term.

Wang Tianxing concluded, “The capital bubble a few years ago pushed autonomous driving to its current position. The supporting facilities required for the full-scenario industrialization of L5 autonomous driving are far from mature, and are currently limited to specific simple point-to-point scenarios.”

"One is that the technical support is immature. L4 and L5 require extremely strong perception capabilities. I personally think that pure vision solutions cannot meet the requirements of L4 and L5 perception capabilities and require Lidar perception assistance, but the current laser cost is expensive. The second is infrastructure facilities. Immature, including V2X (the exchange of information between vehicles and the outside world), roads, 5G, and infrastructure are too expensive.” Wang Tianxing further stated.

The infrastructure cost required for the development of autonomous driving can be seen from the investment in my country’s first fully automated “super highway” Hangzhou-Shaoyong Smart Highway. According to media reports, the line will be equipped with a dedicated autonomous driving lane to support the free driving of autonomous vehicles across the line. The total investment is as high as 70.7 billion yuan, and the average construction cost per kilometer exceeds 400 million yuan.

Wei to aspects of "deep dive insight," said the high-level autopilot is a full -zhi energy of the transportation system, including intelligent roads, high-speed communications and intelligent city, etc., rather than relying solely on the car. To achieve L4 or higher level autonomous driving, an intelligent revolution in the entire road traffic environment is definitely needed.

However, it needs to be pointed out that even if the technical and financial difficulties are overcome, the construction of municipal roads in various countries can meet the requirements. Autonomous driving also has considerable resistance at the policy and ethical level. It requires traffic laws and regulations to be able to fully keep up, and consumers must It also takes time to accept high-level autonomous driving.

Even in the United States, where the automobile industry is developed, Tesla's road to fully autonomous driving has not unexpectedly aroused the attention of regulators.

According to foreign media reports, US federal regulators are paying close attention to Tesla's fully automated driving test version, and the US government is adopting a wait-and-see attitude. A spokesperson for the National Highway Traffic Safety Administration pointed out in a statement that they will "closely monitor this new technology and will not hesitate to take action to protect the public from unreasonable safety risks."

At present, L3 level autonomous driving has been abandoned by some companies due to the inability to clearly divide the responsibility of the accident and other reasons, and directly start to study L4. For safety and other considerations, Robotaxi's tests are still facing scenario restrictions on a global scale, and most of them still need to be equipped with safety officers. But as Wang Tianxing said, "L4 doesn't make any sense without removing the security guard."

"To truly realize L4 and L5 level autonomous driving, it is not something that a car company can do. It also involves the legal and traffic regulations. For example, now China only has a few designated areas in Beijing, Shanghai, Guangzhou and Shenzhen that can carry out L4 and above. Autopilot, more areas are not allowed to leave the steering wheel to drive automatically. Therefore, it is not that the technology is enough, it depends more on whether the relevant laws and regulations are perfect. It is a relatively long process." Chi Zhang Said to "deep dive insight".

5 Conclusion

Although clearly aware that the development of autonomous driving faces many difficulties, Zhang Chi is still optimistic about the development of the track.

In his opinion, it takes a long time for autonomous driving to fully mature and self-profit. The development will not be particularly fast, but once it gets up quickly, it will become standard. "There are more than 200 million cars in the Chinese market. If they are equipped with autonomous driving, the market space will be very large."

In the face of the vast market, even if there are many difficulties, autopilot players are still willing to participate in this protracted battle. Perhaps this is exactly where the charm of investment and technology lies.