Dare to ask where the future is on the road?

Picture source@视 图片 图片

Text ｜ AI Industry Research Center

Today, if you take a taxi or bus without a driver in Shenzhen, you don't have to be surprised.

On August 1st, the first domestic autonomous driving legislation- The Regulations on the Management of the Shenzhen Special Economic Zone Intelligent Connected Automobile Management "began to take effect, which means that autonomous vehicles can be on the road. Immediately after August 8, the Ministry of Communications issued the "Guide to Automobile Transportation Safety Service (Trial)" (draft for comments) to publicly solicit opinions to encourage and standardize the application of autonomous vehicles.

The successive promulgation of the Regulations and Guidelines shows the state's attitude of support from the legal level. This shows that the era of autonomous driving is really coming!

The second half of the autonomous driving starts, and each subdivision track is blooming

As a product of the deep integration of new generation of information technology such as the automotive industry and artificial intelligence, the Internet of Things, and high -performance computing, autonomous driving is an important development direction in the field of global transportation smart travel.

At present, according to the SAE standard autonomous driving, it is divided into six levels of L0 ~ L5. In popular terms, L1 ~ L3 belongs to auxiliary driving, which is equivalent to liberating the driver's hands and eyes; when it reaches L4 ~ L5, the vehicle is completely controlled by the system. At this time, the driver's brain is liberated.

Autonomous driving grading

From the figure, from L3, the system began to take over the vehicle driving, and the driving leadership has completed the human -machine swap. Therefore, L3 is an important watershed for the development of autonomous driving.

From the perspective of specific practice, there are two main R & D roads for autonomous vehicles. One is the "progressive" road that most companies adopted from L1 ~ L5. The other is the "insert" road represented by technology companies, that is, skipping L1 ~ L3, starting directly from level L4, and eventually reaching L5.

According to the track, the existing products are mainly divided into two categories: passenger cars and special vehicles. The downstream application scenarios are mainly concentrated in Volkswagen (including private cars, taxis, bus buses), logistics transportation (trunk lines (trunk lines) (trunk lines) Logistics, city with logistics, end logistics), special use (mine, port, sanitation).

List of major domestic autonomous driving brand information

From a technical point of view, Baidu Apollo and Yuan Rongqi are leading in small passenger car tracks. For example, Apollo directly integrates various sensors such as radar on the vehicle. Without secondary transformation of the vehicle, it can complete control, decision -making, and avoiding obstacles. Yuan Rong Qixing simply removed the "set -top box", incorporating all the car radar into the body, making the appearance look almost the same as that of ordinary cars.

Shenlan Technology is unique on the large bus track. It has developed the 12 -meter panda smart bus equipped with smart driving, smart cabin and other functions. It is the only domestic autonomous driving test in 5 cities in 5 cities in Shanghai, Shenzhen, Guangzhou, Changsha and Wuhan. Commercial vehicles announced with the commercial and information of the Ministry of Industry and Information Technology, and successively won the urban bus procurement project in Shanghai, Tianjin, and Changzhou.

In the field of logistics, AUTOX is the leader. It is the first domestic startup to obtain the California road testing license. It is also the only company in the world to obtain a dual -operating license license for unmanned rental and unmanned goods transportation. Certification.

Public information shows that starting from July and August last year, the autonomous driving rental of Baidu Apollo and Yuan Rongqixing started roads and operation tests in Shenzhen; and Shenlan Panda Smart Bus, which started in May the previous year in the Pingshan Demonstration Zone in Shenzhen Essence

With the release of policies and the improvement of the law, the second half of the domestic autonomous driving that focuses on "commercialization" has begun. Next, the entire industry will inevitably develop well -spray development, and large -scale commercialization also means that it is just around the corner. Those autonomous driving companies that broadcast the "technical landing" seeds early will also be worthy of this opportunity to taste the fruits of commercialization.

AI Gold Circuit, Multi -City whistle running on the autonomous driving

At present, with the popularity of 5G equipment and vehicle -road collaboration facilities, autonomous driving ushered in the best opportunity for leapfrog development. In addition to Shenzhen, multiple cities across the country have opened a demonstration of autonomous driving road testing, which will greatly accelerate the process of industrialization, large -scale, and commercialization of autonomous driving.

Beijing is the first city in China to measure the green light on the public road to the autonomous road, and the first pilot pilot of the driving vehicle unmanned operation. Shanghai took the lead in setting up a national -level industrial demonstration group, a total of 25 enterprises issued road test permits, and opened 615 roads for testing roads, with more than 12,000 scenes. Guangzhou recognizes other regions of intelligent connected car road test permits, conducts ROBOTAXI technical verification in batches, and carry out road testing in the main road of the central urban area.

According to the statistics of the Huaxia Times, in addition to the Beijing, Shanghai, Guangzhou and Shenzhen, more than 40 cities including Changsha, Wuhan, Suzhou, and Chongqing have issued policies and management measures to actively carry out road testing, manned testing, and demonstration operations of autonomous driving.

The implementation of autonomous driving cars in major cities in China

It can be seen from the figure that travel and logistics are the two most popular application scenarios in China at present, and there is a trillion -scale emerging market hidden behind it. Multi -cities take autonomous driving as "knocking on the stepping tiles", and strive to knock on the door of this market.

Autonomous vehicle, "banknote fragment" on the wheels

In the past two years, various enterprises have involved in autonomous driving. Whether it is BYD, Tesla, Baidu, Google in the Internet, or science and technology giants such as Huawei and Apple, they have begun to actively lay out and invest. However, since 2020, the experience of autonomous driving can be described as "two days of ice and fire." On the one hand, national policies are constantly favorable, the capital market continues to be optimistic, and corporate financing is better. Some companies have received several large amounts of financing within a year.

At present, the most easy -to -financing of the track is unattended. For example, Xiaoma Zhixing, financing of over 720 million U.S. dollars in 2020, and harvesting another $ 100 million in 2021 to add financing. Wen Yuan Zhixing, led by Yutong Group in 2020, led a $ 200 million in Yutong Group, and then completed three consecutive rounds of financing for a total of 440 million US dollars.

On the other hand, if you do n’t talk about investment and financing, just look at the company's own operating data, the financial situation is bad. According to Pitchbook research data, start -up companies engaged in the autonomous driving industry have a monthly operating cost of $ 1.6 million (equivalent to 10.51 million yuan). In general, the entire industry is basically presenting a strange circle of "more money is mellow and faster and faster".

How fast is the speed of autonomous driving? The Uber's autonomous driving department has burned 2.5 billion US dollars in 5 years; Waymo spends more than $ 1 billion each year. It is Baidu that is most willing to "burn money" in China. According to financial report data, it has invested 20 billion yuan in 2020. Han Xu, CEO of Wenyuan Zhixing, also told the media during the interview: The company has maintained the speed of spending 100 million US dollars per year.

In fact, since the date of birth of autonomous driving, it is a business that is "difficulty in investing in effectiveness, difficult technical landing, and difficult to commercialize", and it is the kind of "only seeing input and no output". How long of money and burning are not even the standard.

Take ROBOTAXI, which has been commercialized, and according to Didi public data, in daily operations, the cost of R & D, manpower, operation, and maintenance of each car is more than 1 million yuan.

Compared with huge costs, revenue can be ignored. According to Waymo executives, it was revealed to the media: In 2020, the US Phoenix Corporation can receive 1,500 businesses per month, and its revenue is only over 10,000 US dollars. In China, although in many cities such as Beijing, Shanghai, Guangzhou and Shenzhen, Baidu Apollo rental, Shenlan Smart Bus, Yuan Rongqixing Rental and other commercial operations have begun, but they all rely on corporate burning money subsidies.

As a industry person said: "Autonomous driving is not a hundred meters, but a marathon. Running on this track is patience and endurance, not all companies can run to the end." It can be seen that autonomous driving is far away to live the kind of life that does not rely on burning money.

Where did the money spent?

Why do you burn money for autonomous driving? Talent, data, hardware, etc. are all reasons.

The importance of talents is beyond words. Autonomous driving is cutting -edge technology, involving AI, computers, big data, mobile Internet and other information technology fields. A large number of high -tech talents are required to participate, otherwise they will not be able to play at all. Right now, global autonomous driving engineers are in a serious shortage state.

Besides data, in order to correct algorithms, build models, verify technical security, and analyze road scenarios, companies need to perform a lot of road testing to collect enough data. Only L3 is autonomous driving, and the road test of 20 million kilometers is required, and L4 is a horrible 1 billion kilometer. Among them, the labor cost of road testing technicians, coupled with the fuel costs, maintenance, and insurance of the vehicle, the expenses are highly high.

In addition, the cost of hardware, the laser, camera, radar and other sensors on the vehicle itself and the body, and various instrument display instruments, the cost of a self -driving car is comparable to sports cars.

In particular, the entire industry has not yet formed a reasonable profit model. Regardless of the Robotaxi of Baidu Apollo and Yuan Rongqixing, or the bus buses of Shenlan Technology and Qingzhou Zhixing, they can only be regarded as commercialized attempts. Compared with huge operating costs, The income obviously "cup of water".

At this stage, the reason why all parties in the industry spend money on autonomous driving without blinking. In essence, no one is willing to miss the big "money" of smart transportation in the future.

Master once pointed out the importance of autonomous driving to Tesla: "Whether it can solve the problem of autonomous driving, it directly determines whether Tesla's value is high or almost worthless in the future."

The current autonomous driving is like a dormant volcano, which has gathered pressure for hundreds of years. Once eruption is erupted, huge energy will erupt.

If you want to burn less money for autonomous driving, companies cannot just eat "soft" rice

In general, the reason why autonomous driving attracts various types of enterprises to enter the bureau is that it can reduce costs, improve efficiency and safety, while creating a new business model, so that enterprises can get new "banknote capacity".

Although the "banknote ability" is seductive, the development of autonomous driving at present has to be maintained by losses. At present, there are few autonomous driving enterprises that can be "burned less and do more". Investigating the root cause, the problem lies with the majority of autonomous driving companies that did not set their positions, causing burning money to burn too much, and even those who should not be burned.

As we all know, either car company or technology company. In terms of capabilities, car companies have strong hardware (vehicle) design capabilities, weak software (autonomous driving system) development capabilities, and technology companies are opposite, while autonomous cars are a product of hard and hard combination. Therefore, the car company follows the "hard+soft" route. To put it plainly, it is "equipped with a autonomous driving system for the car", but the technology company follows "soft+hard", that is, "loading the autonomous driving system to the car to the car superior". "Hard+soft" focuses on reliable, "soft+hard" emphasizes performance, and the dominance of the two is completely different. Although essentially speaking, no matter who dominates, you can get autonomous cars, but the amount and cost of the two are different. Because, from the perspective of industrial maturity, the vehicle design maturity is high, and the autonomous driving system is far from mature. How can an unfamiliar software be dominated by a mature hardware joint development? If this is done, isn't the risk higher?

Obviously, if you want to avoid the current strange circle of "burning too much money" for autonomous driving, the key is to make these technology companies "soft rice hard", that is, not only system development, but also the whole vehicle design. Corporate work to help you do it together. For example, Baidu Apollo's "radish fast run" is designed with Beijing Automobile; Robotaxi of Yuan Rong Qixing is customized for it for it; and the 12 -meter panda smart bus of Shenlan Technology, the entire vehicle is even from Shenlan Designed and developed by yourself.

At the same time in the development of autonomous driving systems and the design of the vehicle, not only avoids the embarrassment of software suppliers, but also greatly reduces the cost of software and hardware compatibility, trial error, adaptation and testing.

Therefore, under the premise of current autonomous driving is still in the stage of pure burning money, we must do "do more things and burn less money". We must abandon the concept of minding to do systematically and follow the path of "hard and hard and hard work+integration research and development" in order to in order Get the effectiveness of "cost reduction and efficiency".

All in all, the future of autonomous driving is still long, and there are many uncertain factors. If technology companies still cannot change the old R & D model, heavy technology, light integration, then autonomous driving will be a one who can not see the end point. Destroyer.

- END -