Autonomous driving is the same as the dreamers, and the companions are far away

Author | Li Dongmei

Interview guest | Yang Wenli, founder of Lingjun Technology

In 2013, the first year of Baidu's autonomous driving officially arrived, and Baidu began to recruit AI talents around the world, and successively established a deep learning research institute, a big data laboratory, and an artificial intelligence laboratory. At the end of 2015, Baidu announced the official establishment of the Autonomous Driving Division (L4), which is also considered to be the first year of domestic autonomous driving.

In the blink of an eye, 8 years have passed, as a pioneer in the field of domestic automatic driving, Baidu's autonomous driving division has not been profitable so far, and even at the end of last year, some media reported that Baidu's driverless department laid off employees in a large area. However, although the head company has lost half of its teeth when it gnaws down the hard bone of autonomous driving, there is still no shortage of fresh blood on this track.

Yang Wenli, the founder of Lingjun Technology, is a bettor who has always been optimistic about the track of automatic driving.

The road to entrepreneurship hurts and is happy

I believe that pain and happiness are the happiest feelings for every tech entrepreneur.

In 2006, Yang Wenli went to Pennsylvania State University in the United States for a doctorate after graduating from Tsinghua Automation, and many international students in the United States, it was natural to join an American company after graduation, and after completing his postdoctoral studies in 2009, Yang Wenli entered the Western Digital Hard Disk Company as the chief architecture engineer. At that time, Google had just started Waymo, a self-driving car company, and the world's self-driving track was still deserted.

At that time, Yang Wenli did not have a very clear plan for entrepreneurship, and automatic driving was relatively far away for him and the domestic market.

But Yang Wenli, who was working at Western Digital at the time, still vaguely felt that the Chinese market on the other side of the ocean was undergoing earth-shaking changes... With the rise of the Internet, 5G, blockchain, and artificial intelligence technologies, autonomous driving has begun to enter people's vision.

After Google founded Waymo, Ford, Mercedes-Benz, BMW and other OEMs have also begun to develop autonomous driving technology, trying to force the unmanned car market.

Baidu, which has won a world in the Internet "fight", naturally has a very high technical foresight. In 2013, Baidu officially announced the establishment of the Deep Learning Research Institute (IDL), with Robin Li personally serving as the dean and Yu Kai as the vice president, and recruiting AI talents around the world. Yang Wenli is one of the talents who was "targeted" by Yu Kai.

In 2014, Yu Kai recalled Yang Wenli from the United States to China and put him in charge of the decision-making, planning, control and simulation modules in Baidu's autonomous driving department. It is this opportunity that gives Yang Wenli the opportunity to experience the development of Baidu's Apollo L4 autonomous driving system infrastructure from 0 to 1.

This experience in Baidu drew a strong stroke in Yang Wenli's career, he lamented that Baidu has done such a feat to lead the domestic autonomous driving wave, but also saw that the learning ability and innovation ability of the Chinese research and development team are actually not worse than the United States, and we can do as well as them in automatic driving technology, or even better.

The sea is wide with fish leaping, and the sky is high for birds to fly. Although Baidu is a good company, independent development has better flexibility, more autonomy and control, and both teams and individuals have more room to play their abilities. As a result, Yang Wenli bred the idea of starting his own business.

"If we can bring automatic driving into life, we can solve the problem of travel for the elderly to the greatest extent, in fact, the corresponding problems are the labor shortage and traffic safety problems after the aging society." It just so happens that my profession can solve these social pain points, and self-employment can achieve the results you want faster. ”

Yang Wenli said: "The initial decision to start a business began with ideals, feelings, and even passions and impulses. But whether it can be implemented or not also requires an objective analysis of factors such as the team, the market, technical capabilities, and industry prospects. Our founding team's own technical capabilities, industry position, experience accumulation, financial capabilities and other factors are enough to support this small start-up team to achieve the common ideal. 'Bold guessing, careful verification' is the basic quality of good scientists and engineers. ”

In this way, in December 2016, after considering various factors and confirming that the time was ripe, Yang Wenli left Baidu and established Lingjun Technology on his own.

Yang Wenli admitted that when he was in school, he was more snoozing, and was jokingly called "sleepygod - sleeping god" on Tsinghua's Mizuki Tsinghua BBS, and later after he started his own business, he kind of lived up to the name. The daily sleep time is greatly compressed. Did you ask him about the pain of such an experience? Of course, the pain, every day is full of strategy, talent, capital, business, technology, cost... But it's more about happiness, he said, "and I believe that pain and happiness are the happiest feelings for every tech entrepreneur." ”

"Five years to sharpen a sword, slow is fast"

Autonomous driving has far-reaching social value, so it makes sense and valuable for us to do this.

After Baidu set off a wave of autonomous driving, domestic automobile OEMs, technology giants and Internet upstarts have entered the game to share a piece of cake.

The blue ocean of autonomous driving is favored because from the industrial, capital and national levels, autonomous driving has fertile soil for its unlimited growth.

From the perspective of industry-driven, the era of traditional manufacturing leading the economy is still past, the era of real estate leading economic development has passed, the innovation time of rapid growth of the Internet has also passed, under the premise of global industrial restructuring and world economic integration, hard technology will become the engine and power to lead the world economy in the next fifty years, and smart cars are the most core plate in hard technology.

From the perspective of capital investment, in 2021, the market size of intelligent driving assistance systems in urban areas will be about 5.8 billion yuan; the market size of high-speed intelligent driving assistance systems will be about 24.7 billion yuan; the market size of intelligent parking assistance systems will be about 13.7 billion yuan; the scale of the autonomous driving market in mining areas will be about 2.1 billion yuan; the scale of the port automatic driving market will be about 200 million yuan, and the total cumulative investment and financing of the primary and secondary markets of automobile intelligence will reach 159.19 billion yuan.

From the perspective of national innovation, in the era of traditional fuel vehicles, Western economies such as the United States, Japan, Germany, France, and South Korea have mastered the core technology, and intelligent new energy vehicles have given China the opportunity to overtake in curves.

Yang Wenli admitted that he was lucky to catch up with such an era. Fortunately, the founding team members who joined him were all friends and colleagues who were optimistic about the field of autonomous driving, and a group of like-minded people working together for the same goal was the first step to get things done.

After the establishment of Lingjun Technology, it did not rush to launch products to open the market, nor did it show its prominence in front of capital, but focused on polishing technology and concentrating on research and development of products, which is a bit like the integration of "Mobieye's R&D personality" + "Tesla's system engineering thinking" + "Waymo algorithm capabilities".

In terms of autonomous driving technology, perception technology is relatively mature, deep learning, laser, vision, etc. have mature models and algorithms, and the capabilities of each family are not much different. The core and most difficult technology is how to game with human drivers, that is, autonomous driving algorithms need to predict human behavior, and then decide their own behavior and put it into action. Human behavior is unpredictable, because there are many influencing factors, and the decision-making of automatic driving and the game with human drivers will directly affect the final performance and function realization of the vehicle, which is the most differentiated core technology point between different product and technical teams.

Since its inception, Lingjun Technology has focused on the entire industry, identified its own advantages and positioning, under the premise that the cost of autonomous driving-related sensors is high and the perception and control computing ecosystem needs to be improved, it has accurately focused on the advanced autonomous driving passenger cars with the goal of "mass production form", and technically created the first and second generation L4 level autonomous driving prototypes in five years with the engineering design concept of "central decision-making level first", and launched a simulation test platform suitable for domestic autonomous vehicle research and development.

In 2021, after five years of precipitation, Lingjun Technology will reduce the dimension of L4 level automatic driving technology in its two products - "Xiao Ant Ju" series of park unmanned minibuses and smart network buses, and officially join the low-speed segmentation track in the automatic driving industry, forming a set of "technology-product-market" commercialization closed loop.

Before 2025, Lingjun Technology will focus on the field of autonomous buses and regional logistics, and after 2025, it will return to the commercial battlefield of autonomous passenger cars.

Yang Wenli said that he himself has always believed in a proverb "slow is fast", which is a sentence from the "Tao Te Ching", that is, to precipitate down, in order to develop faster and go further.

"Autonomous driving belongs to the emerging high-tech industry, but self-driving cars are industrial products, which will be applied to people's lives and social service systems in the near future, and the product cycle of the industrial system is not as fast as everyone thinks."

"Fighting" entered the second half, and the strength was strong to survive

Since the development of the industry, the days when you can get financing by relying on a single paper PPT have passed.

After several years of rapid development, startups on the autonomous driving track have made great progress in the landing application stage, and the market size is also expanding. In 2021, the market size of autonomous buses will be 300 billion, the size of the autonomous urban distribution market will be 500 billion, and the market size of autonomous passenger cars will be 2.5 trillion. Compared to such a large market, there are not many players on the track now.

Even so, the competition is still cruel, because the industry has developed so far, and the days when they could get financing by relying on a single paper PPT have passed. The game of autonomous driving technology has entered the second half, and it is time for each family to show its "housekeeping skills".

Yang Wenli said that Lingjun Technology has been deeply engaged in the field of automatic driving for many years and has achieved good results in the following aspects:

First, the technical ability, technical results and decision-making planning algorithm ability of the first echelon are more prominent;

Second, the simulation platform independently developed by the core R&D tool chain, the self-developed high-precision map tool, and the L4 autonomous driving passenger car technology stack are relatively perfect;

Third, the data iteration and data rumination system is sound;

Fourth, the investment in lean research and development is low;

Fifth, the invalid cost is wasted and the historical burden is less;

Sixth, bringing together talents from software to hardware, from electrical architecture to system engineering;

Yang Wenli said frankly, "The process of accumulating strength in all aspects is easy to say, but this period is full of dangers and obstacles. In the most difficult period of the company, there were more than a dozen companies in the industry waiting for us to consume the funds to buy us at a low price, there were car companies, suppliers, Internet companies, and technology companies, just like a group of vultures staring around the dying child, waiting for him to fall, swallowing him. Fortunately, Lingjun Technology survived. ”

Autonomous driving is the same as the dreamers, and the companions are far away

Division of labor and cooperation is the essence of modern industrial large-scale production.

In 2022, nearly 13 years have passed since Dr. Wenli Yang graduated. When mentioning whether he regrets entering the field of automatic driving, Yang Wenli said:

"I never regretted it. If I had just graduated with a Ph.D., I might have chosen to get involved in the autonomous driving industry right away. For the autonomous driving industry, 2021 to 2025 will be the best five years, which is also an important reason why Lingjun Technology chose to restart in 2021. Autonomous driving is a comprehensive discipline, not just theoretical research, nor just software engineering. From searching for papers, to pushing formulas, to writing code, to real road tests, and finally forming products, how to upgrade what I know and learn into cutting-edge technology, and finally forming products, is the 'precious thing like a diamond in this era' that I harvested at an autonomous driving company. ”

After entering the field of automatic driving for many years, Yang Wenli is full of love for this industry, but in fact, there are many voices in the outside world that are not optimistic about this industry. Self-driving head players Baidu Apollo and Google Waymo have not yet achieved profitability, and other companies are even more difficult.

Yang Wenli frankly said that there are many players driving this track, but there are not many profiteers, mainly because of the industry development cycle. The fast landing is the misleading of the market by industry participants using Internet thinking financing. At present, domestic automatic driving has experienced more than 8 years of development, in addition to easy-to-use, there are several that can achieve profitability, but this is the development cycle that the industrial system should have. Ten years is the productization cycle of intelligent industrial products, if anyone can achieve breakeven between 2023 and 2024, and the revenue is sustainable, who can really lead the future.

According to Yang Wenli's judgment, 2014-2020 is the maturity period of the core supply chain of autonomous driving, small-scale commercialization will be realized in 2021-2025, 2026-2030 is the period of commercial acceleration + industry pattern reshaping, and 2031-2050 will be the twenty years that smart cars will lead the world economy.

In the field of dynamic driving, many traditional enterprises and technology companies are deeply bound, because the vehicle chassis is the core carrier of automatic driving, and without the vehicle, automatic driving cannot be talked about. There is both division of labor and cooperation between car companies and technology companies, and the classic is the relationship between mutual suppliers.

Technology companies buy vehicles from depots, which transform vehicles and provide technical support; depots obtain products, solutions and technical support for autonomous driving technology from technology companies. However, there is still a key element missing from this - the scene. Yang Wenli said, "The depot-technology company-scene side is the best iron triangle relationship for automatic driving, such as Xiamen Golden Dragon - Lingjun Technology - Ganzhou Government/ Suzhou Government, Dongfeng Yuexiang - Lingjun Technology - Wuhan Government, etc." ”

In order to develop autonomous driving technology better and faster, Yang Wenli said: "In addition to the cooperation between technology companies and car manufacturers, there are chip manufacturers, sensor manufacturers, controller manufacturers, system integrators, etc. Suppliers of all aspects of automatic driving need to be twisted into a rope to form an ecology in order to climb the Everest of automatic driving." ”

Autonomous driving is the same as the dreamers, and the companions are far away.

Leifeng Network