In a recent public speech, Baidu founder Robin Li said of the development of autonomous driving: "Baidu and Tesla are both engineering, and Google's Waymo is scientific thinking."
Robin Li actually changed the terms of how to commercialize, and wanted to express how the emerging autonomous driving industry makes money, and whether it should make money first or later.
As a pioneer in China's autonomous driving industry, Baidu naturally knows what the biggest dilemma the industry is currently encountering.
In the past three years, in the big game of Baidu autonomous driving, the weight of commercialization is increasing day by day. So, can a new business path different from Google Waymo really complete the game?
The crossroads of commerce
The global autonomous driving industry has been following the "bike intelligence" route dominated by Silicon Valley for a long time.
Engineers define self-driving technology as a closed loop from perception, decision-making and path planning to control execution. By collecting scene data on real roads and training algorithm models at each node, it is hoped to achieve absolutely safe autonomous driving.
Around 2014, a number of domestic self-driving entrepreneurs born out of Google Waymo and Baidu's unmanned car team followed suit.
After a long iteration, the status quo shows that you can't gather all kinds of strange and extreme scenes in the real world where the camera is obscured by leaves, backlight, heavy rain, white trucks, pedestrians in cosplay Pikachu, etc., and it seems that there is always an inexhaustible space, so it is difficult to ensure the safety of automatic driving.
Without safety prerequisites, autonomous driving is difficult to land on a large scale, which also triggered a sharp decline in Waymo's valuation and team turmoil.
The "vehicle-to-road collaboration" technology has long been discussed in academia, but the real application of it to the application and commercialization of the autonomous driving industry is around 2019.
The premise for the implementation of this technology is that the state has promoted the construction of new infrastructure represented by 5G and smart transportation upgrades.
This technology introduces intelligent perception and computing equipment on the roadside and in the cloud, which can communicate with the car, and guide and control the vehicle, and the car and the road cooperate, ultimately enabling the automatic driving system to achieve higher safety and effectively solve the inexhaustible long-tail problem of the above bicycles. In addition, the equipment on the side of the road can share the perception and calculation work on the car, which also reduces the cost of autonomous vehicles.
The autonomous driving industry is beginning to shift. At the end of 2019, Li Zhenyu, the head of Baidu Apollo, announced at the first apollo ecological conference that "bicycle intelligence + vehicle-road collaboration" will go hand in hand, and put vehicle-road collaboration to the strategic height of driving side by side with automatic driving.
Robin Li later pointed out at the World Forum on Science and Technology and Development held in November, "We have never relied solely on bicycle intelligence. The improvement of infrastructure is also very important, vehicle-road collaboration can greatly improve the safety of unmanned vehicles, when the safety of unmanned vehicles greatly exceeds that of manned vehicles, the opportunity for large-scale commercialization will come. ”
In 2019, Zhu Lei, founder and CEO of another autonomous driving company, Mushroom Car Union, also said at a meeting that "bicycle intelligence and vehicle-road collaboration are not two isolated technical routes, and the integration of the two is the way out for the commercialization of autonomous driving."
A year later, in August 2020, Baidu Apollo successively won the "new infrastructure" project of intelligent transportation for autonomous driving and vehicle-road synergy in Guangzhou and Changsha, of which the amount of the Guangzhou project was nearly 460 million yuan.
When the contract was signed, Robin Li personally flew to Guangzhou to attend the ceremony. According to statistics, as of now, Baidu has announced more than 20 cities with tens of millions of smart transportation orders, with an average winning amount of about 160 million yuan.
In October 2019, Mushroom Car Union completed the first "vehicle-road collaboration" 5G commercial demonstration project in China in Beijing. Similar to Baidu, the company has signed contracts with Hunan Hengyang and Henan Hebi, with project amounts of 500 million yuan and 300 million yuan respectively.
Up to now, according to incomplete statistics, in the past two years, only Baidu and Mushroom Car Union have signed smart transportation projects with autonomous driving as the core of the amount of more than 1.52 billion yuan and 1.16 billion yuan respectively.
Most players are in a dilemma
Autonomous driving has found a way out of large-scale landing. As the promoter of the "bicycle intelligence + vehicle-road collaboration" program, companies such as Baidu and Mushroom Car Union have begun to taste the sweetness of the autonomous driving industry.
In the past two years, the two companies have gradually developed the "bicycle intelligence + vehicle-road collaboration" program into a relatively complete technical implementation system.
Baidu calls it "ACE Intelligent Transportation Engine", the core of which is the integration of its "vehicle-road cloud map" capability; mushroom car-to-body is called "vehicle-road cloud integration" full-stack technology solution.
Both are trying to become technology system integrators in the field of autonomous driving, providing overall solutions for smart transportation and autonomous driving to government and commercial big customers.
It is in these two years that autonomous driving has gradually increased from the "money shortage" that began in 2018, and since the middle of 2020, it has ushered in a new round of financing frenzy in history, and this tide has continued to this day.
The high point occurred in the first half of this year, when self-driving truck company Tucson Future landed on the NASDAQ as the world's first autonomous driving stock and quickly reached a market value of tens of billions of DOLLARs. Before and after its listing, according to incomplete statistics, since the beginning of 2021 alone, the total investment and financing in China's autonomous driving field has reached 42 billion yuan.
Now it seems that this creates a false illusion for the self-driving industry. The rapid listing of the "Tucson model" in the United States has become the path that most entrepreneurs in the autonomous driving industry aspire to, and even the ultimate goal. They know that it is not easy to commercialize autonomous driving, it is difficult to achieve self-hematopoiesis, and instead hope to maintain the survival and continuation of the company through public financing.
Moreover, listing in the United States, where conditions are more relaxed, has almost become the only choice for self-driving startups with high growth expectations but difficult to cash in in the short term.
In July, affected by the Didi incident, the Chinese and American governments have strengthened the supervision of Chinese companies listed in the United States, and the door to IPO in the United States has almost closed, and companies that have withdrawn their listing plans have emerged one after another. This may be a fatal blow to many Chinese self-driving startups.
In a dilemma, most autonomous driving companies can only continue to show technology, tell stories, and speculate on valuations, but rarely talk about the life-and-death topics of project landing, commercialization, and revenue status.
The reality is that Tucson, the only self-driving truck company currently listed to disclose financial reports, will only have a revenue of 10 million yuan in 2020. This is also a manifestation of the U.S. market, where truck drivers pay higher wages and the threshold for technological substitution is lower.
At a conference on the autonomous driving industry at the end of September, when asked by the host about the above topic, several previous star entrepreneurs in the industry were invariably embarrassed. Some people even retorted, "I think people should be ashamed of making money quickly," one founder said.
In contrast, only a few companies such as Baidu and Mushroom Car Union continue to talk about the commercialization of autonomous driving and how to land on different occasions. They have become a minority in the current self-driving industry.
The industry landscape will change
With the successive landing of a number of domestic "vehicle-road collaboration" projects after 2020, this model led by Baidu has also brought subtle adjustments to the industry pattern.
In September, the "Intelligent Connected Vehicle Industry Development Report (Shunyi Index 2021)" released by the China Electronics Information Industry Development Research Institute showed that Baidu Apollo and Mushroom Car Union ranked in the top two, forming the "first echelon" of the current autonomous driving industry with a comprehensive performance of 87.9 points and 80.4 points, respectively.
The report believes that the autonomous driving industry has entered a reshuffle period, and landing capabilities and business models are the key at this stage.
From the above-mentioned automatic driving rankings, it is not difficult to see that there is a simple and clear criterion that separates the above two companies from other players, that is, whether they have a complete set of autonomous driving landing solutions and business practice projects of "bicycle intelligence + vehicle-road collaboration".
At the recent Peking University New Engineering International Forum, in the eyes of the outside world, Robin Li proposed that Baidu Autopilot and Tesla are both "engineering thinking", but with the previous background, we can understand why the founder of this technology-extreme company began to become pragmatic.
His explanation is that scientific thinking is a step to the sky, and engineering thinking is a step-by-step, first in some scenarios to achieve automatic driving. Tesla has adopted such a method, it first does L2, follows the car on the highway, changes lanes, and realizes automatic parking in the parking lot.
"This way has succeeded", Robin Li said, the market is very accepting, the car is selling, every day countless people drive Tesla cars for them to collect all kinds of data for free, Tesla thus has an unparalleled super-large-scale autopilot-related data, so investors think that Tesla is more likely to make unmanned driving.
Interestingly, the self-driving industry has mixed feelings about Tesla. Many people think that Tesla's low-cost, purely visual self-driving technology is a reduced version, at most a transitional plan, and scorn it, regard it as an outlier; on the other hand, it is red in the eyes and the heart is not a taste for its stock price to sing all the way and the vehicle is selling well.
This attitude has also been extended to the "bicycle intelligence + vehicle-road collaboration" scheme proposed by Baidu and a few companies. Many people in the automatic driving industry believe that the "vehicle-road collaboration" is proposed because the "car' automatic driving research and development capabilities are not too hard, and the "road" collaboration is needed.
In fact, both the "technical school" and the "engineering school" are likely to succeed and will encounter great difficulties; and in today's multidisciplinary integration, science and business dance together, it is difficult to distinguish between pure "technical school" and "engineering school", which is not the most important thing.
The bottom layer of "bicycle intelligence + vehicle-road collaboration" is a huge real-time data architecture system, which may create far greater social value than allowing a car to achieve driverlessness, but at the same time, companies such as Baidu and Mushroom Car Union also need to cross greater technical difficulties, make the huge technical system efficient, collaborative, and quickly master the way to do business with the government.
On the other hand, Google's Waymo and its Chinese followers may also be unmoved, achieving a qualitative breakthrough in the single-point technology they believe in, and eventually achieving a "reversal".
The important thing may be that in the current changeable capital environment, autonomous driving companies quickly find ways to make themselves hematopoiesis, forming an enhanced loop of "landing - obtaining data - algorithm iteration - landing". Commercialization is the one in front of the innumerable 0s.
If we continue to shy away from the topic of how autonomous driving makes money, these startups may close or be acquired within the next year or two. The scale of commercialization at this stage will also gradually open up the gap in the technical capabilities of autonomous driving companies.