On October 19, 2016, Hangzhou Xiaoyu, Lou Tiancheng attended a social occasion as a non-technical person for the first time.
At that time, he had the idea of starting his own business, and he seemed to be mentally prepared to transform from a “technical nerd” to an “entrepreneur”. He met many people of all kinds. This time, he is going to attend the annual CEO Summit held by Sequoia China in Hangzhou.
The Shangri-La is full of toasts and people’s voices are noisy. Lou Tiancheng sat across from Shen Nanpeng, who was then the head of Sequoia China. Presumably the atmosphere was subtle and the conversation was tense. Six years later, Lou Tiancheng recalled the scene to Leifeng.com in Pony.ai’s office in San Francisco, USA:
At that time, it was the time when domestic self-driving cars were on fire, and there were many people with little money. It was rumored in the circle that Lou Tiancheng and Peng Jun intended to leave Baidu to start a business. IDG and Sequoia first approached them and expressed their investment intentions. IDG was discussed first by Ding Fei and later by Guo Yihong, while Shen Nanpeng personally talked about Sequoia.
At the summit dinner, Lou Tiancheng and Shen Nanpeng shared the same table. Although both parties tacitly wanted to discuss the investment and cooperation of autonomous driving startups, Shen Nanpeng did not talk about anything related to autonomous driving. Instead, he asked a bunch of irrelevant questions, which made Lou Tiancheng dizzy. But Lou Tiancheng can also answer, giving a good answer.
The dinner party was short. They didn’t have an in-depth conversation that day, Lou Tiancheng went back a little confused. However, after the first round of financing of Pony.ai, IDG and Sequoia invested a total of 15 million US dollars in the same round, and Sequoia took the initiative to add a little more.
Later, Shen Nanpeng told Lou Tiancheng that he just wanted to see how his performance was. Make people dizzy first, and then look at the true nature of people. This is Shen Nanpeng’s consistent technique for testing entrepreneurs.
What he didn’t expect was that Lou Tiancheng was not afraid of being seen through.
That year, Lou Tiancheng was 30 years old. Leifeng.com
01 Entrepreneurship
Regarding Lou Tiancheng’s entrepreneurship, there were actually two different voices in his circle of genius friends back then.
One is surprise. Leifeng.com
Lou Tiancheng and Peng Jun founded Pony.ai at the end of 2016 with a founding team of 7 people. The office was first set up in Fremont, San Francisco, California. But the news spread quickly. Almost everyone in the alumni circle of Tsinghua Computer Science Department and the Chinese programming competition circle who followed Lou Jiaozhu got the news immediately, including Pei Xiaohui, who was teaching at Nanyang Technological University in Singapore at the time.
Bei Xiaohui and Lou Tiancheng have known each other since high school. They are both informatics contestants. They entered the national training team together. After being sent to Tsinghua University, they were classmates again. They studied undergraduate and doctoral degrees together. They have a good friendship. After leaving Tsinghua, although Lou Tiancheng went to Google in the United States, Bei Xiaohui went to Germany and Singapore and stayed in academia, but the two still have contact.
Bei Xiaohui was a little surprised when he heard that Lou Tiancheng started his own business. Because in his impression, Lou Tiancheng has always been an introverted “technical nerd”, who likes to mess around with programming and technology, and is not very keen on socializing. When he was in college, he didn’t even join a decent club.
There is no doubt that starting a business requires strong technical skills. But in Bei Xiaohui’s understanding, besides technology, entrepreneurship requires other abilities, such as judgment on the market, dealing with people, and managing the team. Obviously, Lou Tiancheng did not have the opportunity to show outstanding abilities in these aspects when he was a student.
This surprise is not unique to the genius circle. Leifeng.com
Li Hengyu (now the person in charge of Pony.ai Trucks) also told Leifeng.com that when Pony’s office in Beijing was first established in 2017, many investors visited the company and often asked the same questions: “(Peng Jun and Lou Tiancheng) Can these two founders manage this company well?” They are all technical backgrounds, and everyone has never doubted their technical strength, but they have never stopped questioning whether they can manage a start-up company well, until today .
And another voice in the opposite direction is: he finally started a business.
Similarly, their strong identification is also based on Lou Tiancheng’s excellent programming and technical skills. In 2012, Lou Tiancheng graduated with a Ph.D. one year ahead of schedule. When he joined Google, Quora, and Baidu successively, although they were all good companies, when they were placed on Lou Tiancheng, they were more emotionally confused: “Just look at programming ability. , how many people in these companies are better than him?”
Many people know that Lou Tiancheng is known as “the first person in Chinese programming”. But in fact, very few people outside the circle really understand his achievements and status in the field of Chinese programming. A few events may help to understand one or two:
In 2004, the four-member Chinese team, including Lou Tiancheng, set off for Athens and won four gold medals at the International Olympiad of Informatics (IOI), setting a record for the Chinese team in eight years. Their previous all-gold record was won in 1996 by the Chinese delegation including former Sogou CEO Wang Xiaochuan and current Pinduoduo chairman Chen Lei. IOI is the highest stage of the information competition for middle school students in the world. Every year, the representatives of the Chinese team are selected through school, city, province, and national levels. They are the best middle school students in the field of programming in China that year.
After going to college, Lou Tiancheng is still obsessed with programming. He not only participated in competitions organized between schools (such as ACM-ICPC), but also appeared in programming competitions held by major domestic and foreign companies many times, such as Google’s Code Jam, Facebook’s Hacker Cup, TopCoder, etc., all over the world programming masters, successively won the overall champion of Baidu Star Programming Contest (2005 and 2006), and won the global champion of Google Code Jam (2008 and 2009)…European and American programming In the international competitions where masters gather, no one from China has entered the live competition, and Lou Tiancheng has entered 10 times for a record.
From 2006 to 2017, Lou Tiancheng ranked first in the TopCoder China list for 11 consecutive years without a single day. As a result, although many post-90s and post-00s competition students have never played against Lou Tiancheng, they are familiar with the ID “AC Rush” (the code name Lou Tiancheng participated in the competition) by heart. Just pass eight questions” can still attract many challengers. The title of “Guru” was not given by Lou Tiancheng himself, but by Tang Wenbin (co-founder of Megvii), an insider.
Therefore, when some geeks who follow Lou Tiancheng hear that Lou Tiancheng is going to such and such a big factory, their first reaction is often surprise, because in their opinion, “No one in these companies can control Lou Tiancheng.” Therefore, When they heard that Lou Tiancheng started his own business, they were not surprised. Instead, they felt that “Lou Tiancheng should have started his own business long ago.”
Wu Chenye, currently an assistant professor at the Chinese University of Hong Kong (Shenzhen) School of Science and Technology, is a person who holds the second point of view.
Wu Chenye was one year younger than Lou Tiancheng, and both studied under Yao Qizhi’s school. In Wu Chenye’s view, Lou Tiancheng is a person who likes to challenge the limit of his IQ very much. During their Ph.D. studies, the activities they played together were intellectual games, such as bridge, which can count cards. But even if it is a game, Lou Tiancheng will play it very seriously and very well.
There is nothing to say about scientific research ability and technical level. Regardless of the problem in any field, Lou Tiancheng can always come up with a very good algorithm to solve this problem from the perspective of a computer after looking at it a little. Therefore, Wu Chenye feels that with Lou Tiancheng’s technical level, he can do whatever he wants, “it’s just that he has his eyes on the autopilot track.”
Unexpected and expected, there are two different voices, but these two voices have a common starting point, that is, Lou Tiancheng’s ultimate pursuit of technology. In a sense, in the eyes of the outside world, Loutiancheng and technology have formed an inseparable whole. Loutiancheng is technology, and technology is Loutiancheng.
In fact, in Pony.ai’s Silicon Valley office, Lou Tiancheng also expressed similar pursuits when talking with Leifeng.com. He said without any concealment that when he left Baidu, he couldn’t stand the lengthy declaration and procurement process, and he had to make plans to buy several cables. He couldn’t accept that such trivial things would interfere with his technology research and development. If he starts a business, at least technically he can have the final say.
02 floor general
At the Yunqi Conference in Hangzhou in 2017, Bei Xiaohui met Lou Tiancheng for the first time after starting his business.
Bei Xiaohui recalled afterwards that he was very happy to see Lou Tiancheng that time, because the two hadn’t seen each other for many years; in addition, there was another “surprise”. He did not expect such a big change in his former friend after graduation for many years:
There were many people at the venue. When Bei Xiaohui saw Lou Tiancheng, Lou Tiancheng was huddling in his seat and typing on the computer. Bei Xiaohui went to say hello, Lou Tiancheng looked up and saw that it was Bei Xiaohui, who was also very happy, and the two had a cordial conversation in a corner of the venue. After chatting for a while, when it was Lou Tiancheng’s turn to give a speech, Lou Tiancheng ran up and began to talk about Pony.ai and automatic driving.
Bei Xiaohui’s research direction is theoretical computer, and he doesn’t know much about autonomous driving. But he clearly remembers that Lou Tiancheng gave a wonderful talk on autonomous driving that day, and many people present were ignited, including Bei Xiaohui. Bei Xiaohui in his student days had never seen this side of Lou Tiancheng. Bei Xiaohui is very impressed that a shy and introverted boy has grown into a CTO who can stand on his own in big occasions.
People who are familiar with Lou Tiancheng have this kind of feeling: Lou Tiancheng can speak out with great passion about the things that he is really interested in. Those around him are easily infected by him.
Like many smart people, Lou Tiancheng is a person who is used to self-talk. After working, he found that sometimes some of his thoughts could not be quickly understood by the people around him, so he began to train himself to express in a more understandable way, especially after starting a business.
When he was a student, in order to play programming competitions, Lou Tiancheng often traveled around the world by plane for more than ten hours alone, several times a year. His handwritten competition diary and code are still circulating on the Internet, inspiring many young people. After founding Pony.ai, in order to do autonomous driving, he also flew all over the world, preaching, financing, recruiting talents, and discussing cooperation…
The only change is that Lou Tiancheng is no longer fighting alone, but has found another partner who can work well with him: Peng Jun. It is often rumored that the relationship between Peng Jun and Lou Tiancheng has split, but according to Leifeng.com, Peng Jun, who is also a technical background, is actually highly consistent with Lou Tiancheng’s values in the development of autonomous driving. If others do not understand, they can understand each other. Immediately comprehended the accent, and often sang in unison, “like two psychopaths”.
In 2016, Pony.ai was established in Silicon Valley. From left to right: Zhai Jing, Wang Haojun, Zhou Yun, Peng Jun, Lou Tiancheng
Li Hengyu’s enthusiasm for autonomous driving was ignited by Peng Jun and Lou Tiancheng successively. Peng Jun is good at starting with reality and prospects, and Lou Tiancheng is the passion for conquering autonomous driving technology exuded from his bones.
Li Hengyu is also a Baidu team. At the very beginning, he and Peng Jun were both engaged in advertising search research and development in Baidu Fengchao. Li Hengyu formed a dotted reporting relationship with Peng Jun due to overlapping projects. He did many projects under Peng Jun’s guidance, and thus became close friends. In 2016, Baidu’s autonomous driving department (ADU) was in full swing, expanding its recruitment. Peng Jun switched to automatic driving first, and also wanted to recruit Li Hengyu.
At that time, Li Hengyu felt that autonomous driving was a black technology and it was still far away. Peng Jun told him that autonomous driving is roughly divided into several parts, perception, system, planning, and control. Autonomous driving is not that difficult to achieve. Let go of your guard, and then Peng Jun immediately went straight: “The current automatic driving is the search of more than ten years ago.” This made Li Hengyu very excited, as if he saw that the automatic driving will be like today’s search. Become the next huge technology gathering point and change people’s lives. Immediately go to the boss of the department and say: “I’m going to do autonomous driving, don’t stop me.”
Then, in May 2016, not long after joining Baidu, Lou Tiancheng flew back to China from the United States and gave a speech to the Baidu ADU team in Beijing. At that time, Li Hengyu, like many people in the team, did not have a deep understanding of autonomous driving. In terms of technology, he could only learn a little bit by studying papers. Li Hengyu had never met Lou Tiancheng before, and thought that like many geeks, he must be very good technically, but his speech ability should be average. As a result, in the three-hour speech, Lou Tiancheng finished the whole process, and his speech was logical and clear, and everyone present was full of energy and laughter throughout the whole process.
Li Hengyu said that Lou Tiancheng’s talk about autonomous driving is not purely about technology, but from his own experience and summed up experience. In Lou Tiancheng’s eyes, technology is not a series of cold numbers, but more like a living, independent individual with life. Technology is his opponent as well as his friend. He respects it and loves it.
Lou Tiancheng told Leifeng.com that he has been convinced since 2013 that autonomous driving will be possible, and the rest is just a matter of time. And the basis of his judgment is also technology.
At that time, Lou Tiancheng was working as a search engine at Google. At that time, autonomous driving was still very new in Silicon Valley, but Lou Tiancheng was very interested when he heard it. From a technical point of view, he could not find any reason to refute “autonomous driving cannot be done”. So at the end of 2013, Lou Tiancheng transferred to Google X Lab to do decision-making control. There are four people in the decision-making control group, and Lou Tiancheng is the only Chinese.
Lou Tiancheng actually studied data mining and social networks when he was a Ph.D. However, according to a number of people who had worked with Lou Tiancheng in Baidu’s autonomous driving department, Lou Tiancheng joined Baidu in the spring of 2016, and within half a year he unified the communication interfaces of the then chaotic modules of Baidu’s ADU. This is equivalent to rebuilding each module on the basis of ensuring the performance of the original system. It is very laborious, but Lou Tiancheng completed it in a short time.
When he was at Baidu, Li Hengyu also cooperated with Lou Tiancheng on a project, which was to reconstruct the bottom layer of the perception system and turn the serial system into a parallel system. This is the first attempt in the field of autonomous driving. Li Hengyu served as the project leader, and Lou Tiancheng, then chairman of the technical committee of Baidu’s unmanned driving division, was responsible for checking technical solutions and code reviews. The project was very successful. It is said that this project later became part of Baidu Apollo.
This is only the result of Lou Tiancheng’s work in Baidu for less than a year. This shows his strong learning ability and technical acumen. Stand your ground. After founding Pony.ai, Lou Tiancheng took this idealism of technology to the extreme.
For example, when Pony.ai was first established, he was determined to make L4, which is also the most difficult technical difficulty in the development of autonomous driving. Compared with L2 assisted driving, it means greater investment and longer time. commercial journey. But Lou Tiancheng made it clear from the very beginning that this is his first principle, and if this principle cannot be accepted, then nothing will be discussed.
“I may care about some relatively imaginary things. For example, I will have a share in the development of autonomous driving in China in the future.” Lou Tiancheng told Leifeng.com.
Since 2016, China’s autonomous driving has been hovering between firm belief and doubt. In the first half of 2021, investors and practitioners fell into a frenzy of belief in autonomous driving; but in the blink of an eye, in the second half of the year, everyone began to disbelieve in autonomous driving. From extreme belief to extreme unbelief, there was only three months in between. However, the technical beliefs of Lou Tiancheng and Pony.ai have not changed.
Lou Tiancheng thinks it is normal for the rapid changes in the industry’s views, because autonomous driving is a different experience after all. When people are just scratching the surface, ideas come and go quickly. What motivates him to keep investing is never the enthusiasm and pursuit of the external market, but his inner pursuit: changing the world with technology.
In Pony.ai, Lou Tiancheng is not the only idealist. It is said that on the first day of the establishment of Pony.ai, the founding team of Pony.ai sat together and said “to make autonomous driving everywhere.” To achieve this goal, they believe that only L4 can truly achieve it, which is the consensus of the founding members of Pony. Perhaps also because of the highly consistent values, the founding team and core members of Pony.ai are very stable.
Li Hengyu revealed that among the first wave of founding members who joined Pony.ai in 2016 and 2017, about 90% of them remained in Pony.ai. This level of stability even exceeds that of some mature companies.
03 Utopia
When an idealist possesses strength while having ambition, it will often form a kind of appeal and attract more like-minded people. The same is true for Lou Tiancheng.
As we all know, in the early days of Pony.ai, Lou Tiancheng, relying on his fame and connections in the technology circle, gathered a group of young talents who also pursued autonomous driving or technology, including many top technology competition champions and gold medalists. Later, even if some people left, Pony.ai was still recognized as one of the strongest autonomous driving teams in China.
Autonomous driving is a long-term investment. A number of geeks told Leifeng.com that Lou Tiancheng’s personal charm played a big role when they first joined Pony.ai. They believe that Lou Tiancheng is a person who can accomplish truly lofty goals down-to-earth. Before that, he was able to insist on playing programming competitions for more than ten years. If he transfers this love to autonomous driving, “it is a very trustworthy thing.”
Under Lou Tiancheng’s “lobbying” and “emotion”, Pony Zhixing gathered a group of “death fighters” like him in the early days, Wang Chengu, Zhang Ning, He Xing, Mo Luyi… as the stars Gathering together, Pony Zhixing is no longer just the personal ambition of Lou Tiancheng and Peng Jun, but a stage built by a group of people. Xiao Ma’s growth today is also the result of the shining stars.
Zhang Ning and Mo Luyi, who are currently in charge of the Beijing and Guangzhou R&D centers respectively, joined Pony.ai in very similar ways: They told Leifeng.com that they were initially invited by Lou Tiancheng to take an unmanned test vehicle , and then saw Xiao Ma gathered a group of smart people, amazed that Lou Tiancheng was able to gather this group of “the most difficult people to deal with”, and joined them one after another. In August 2017, Zhang Ning went to Silicon Valley for a test drive. After returning to Canada, he packed his luggage within a month and ran back to Beijing to work hard.
Zhang Ning and Lou Tiancheng also graduated from Tsinghua Yao Class. He said that he had experienced the highest density of talents in his life when he was in Yaoban, surrounded by star-studded and proud people. After graduation, he went to Canada to study and work, but he has always missed that time. Therefore, when he saw the talented team of Pony.ai, he somewhat felt a sense of sympathy among heroes. Mo Luyi is the only female champion of ACM-ICPC in the past 30 years. She is also moved by the idea of ”wanting to work with smart people”. Can do something.
Trust is the first step. However, Lou Tiancheng concluded to Leifeng.com afterwards that to maintain the combat effectiveness of a team, especially a team composed of a group of smart people, it is necessary to accumulate bit by bit. The understanding and judgment of technological development is the foundation. “The most important thing is to lead everyone to victory.”
Pony.ai also gathered a group of people who were the first to engage in autonomous driving in China. He Xing joined Didi in 2015 to research autonomous driving. He was one of Didi’s earliest people to work on autonomous driving. He is now the technical backbone of Pony.ai’s truck business. He said that before he joined Pony.ai, he once doubted whether autonomous driving was a “scam”, but Lou Tiancheng changed his mind.
In 2017, He Xing resigned from Didi and interviewed Xiao Ma. When he was in Didi, He Xing once encountered the bottleneck of technological iteration. Various algorithm modules competed for the same computing resources, which caused the system to remain unstable. It took a lot of time to find a solution. Therefore, He Xing believes that autonomous driving may not be realized on a large scale for a while.
However, Lou Tiancheng raised a point during the interview: autonomous driving has reached the stage of stacking projects layer by layer, just like building a house. This directly answers all the questions that He Xing had previously encountered – because they did not follow the scientific engineering method to develop the project before. In addition, Lou Tiancheng also answered many of his doubts, and he resolutely joined Xiao Ma to participate in building blocks.
He Xing also confessed to Leifeng.com that before working with Lou Tiancheng, he thought that Lou Tiancheng was not a “flexible” person. For example, he could play games for more than ten years. But later, from the perspective of actual work, he found that Lou Tiancheng is a person who can be flexible in the face of different challenges, but he is not willing to opportunistic. “After tearing off all the camouflage, he will follow the path that can reach the other side in a stable and reliable way.”
Behind this “stubbornness” is actually a kind of respect. Lou Tiancheng seems to be very clear that the self-driving Rome cannot be built in a day. How to make the technology stand the baptism of time and the test of landing is more important than reaching the destination faster. When talking with Leifeng.com, Lou Tiancheng also said frankly: he doesn’t care whether Pony.ai is the fastest company to make autonomous driving, because he thinks that not many companies will reach their destination in the end.
After starting his business, Lou Tiancheng no longer spends much time allotted to writing code, but spends more time discussing with business and products. When a manager needs to understand the difficulty of doing things for each team, each team must have a clear direction, and the direction he said must convince everyone. Lou Tiancheng used the concept of “distributed system” to describe his management goals for the team:
“A single strong Master (host) is not enough. You need to empower everyone in the team, clearly state the results you want, and ensure that the team’s goals are consistent with the company’s goals, so that everyone can really go to the goal. go.”
There were a lot of masters in the early autonomous driving track, and the competition was fierce, but Pony.ai stubbornly broke through and became a star company on the track. It is far ahead in business lines such as robotaxi, trucks, and driverless hardware. The proud geniuses gathered in Pony.ai believe that this has a lot to do with Lou Tiancheng’s technical leadership and judgment.
Zhang Ning, Mo Luyi, Wang Haojun (COO of Pony.ai), Li Hengyu and others talked about their experience and feelings since joining Pony. They all used one word: fast.
Zhang Ning returned to Beijing in October 2017. Before that, he knew nothing about autonomous driving. He started with simulation, and within a few months after joining Pony.ai, he participated in the assembly and assembly of Pony.ai’s first unmanned vehicle in China. In Guangzhou Pioneer, unmanned vehicles started running in Nansha, Guangzhou within two months.
Mo Luyi joined the Colts in the summer of 2018. In the beginning, unmanned vehicles only ran on a few roads next to the Nansha City Government, only a few kilometers away. The conditions at that time were very difficult. Pony.ai landed on Lingshan Island in Nansha, Guangzhou. The island was uninhabited. They belonged to the real “pioneers”. It was then that they began to develop Robotaxi—to allow employees to use Robotaxi to commute. In April 2021, in less than three years, Robotaxi will be able to run around the clock in an area of 800 square kilometers in Nansha. “The changes are really beyond imagination.”
Wang Haojun said that with each technological iteration, they are getting closer to their goal. In 2017, their car could run a few laps without hardware problems, but it still needed manual daily maintenance; in 2018, in WAIC, their PonyAlpha generation could last for half a week, running five or six hours a day, and then 8 to 12 hours a day. Now the stability of the system is no longer the same, “it can run for up to ten hours in a row.”
Not only technology, but policy development is also very fast. Li Hengyu is the No. 1 employee of Pony.ai in China, and participated in the whole process of Pony’s landing in Beijing. In 2017, they conducted a road test in Beijing, which was a “black run”, because without policy support, the traffic police could arrest them. In 2018, in order to obtain the first license plate for testing unmanned vehicles in Beijing, they squatted every day at the unmanned vehicle testing ground at the foot of Fenghuangling Mountain for testing and examinations. They squatted for a month before getting the license plate. But today, in the past four years, Beijing has allowed Robotaxi with no driver to receive passengers in Yizhuang.
In the early days of Pony.ai there was a strong “chicken blood” atmosphere, which infected everyone. The efforts of one person may be an example, but the efforts of a group of people describe the effect. They are upside down day and night, immersed in the pleasure of breaking through technical difficulties one by one. All of this is due to Lou Tiancheng’s contribution. Not only technical decisions, but also the spirit of high requirements for technical quality.
Such as the pursuit of excellence. He Xing said that when he first joined Xiaoma, the visualization of Xiaoma’s autonomous driving platform was particularly poor. He “can’t get used to the poor visualization of Loujiazhu’s company”, so he first refactored the visualization platform on a large scale and wrote 20,000 to 30,000 lines of code within a year.
Zhang Ning said that perhaps also under the influence of Lou Tiancheng, Pony.ai formed an atmosphere where “everyone firmly believes that what they do is meaningful and can change society”. Because of the difficulty, they want to challenge more; at the same time, because of the purity, they recognize their efforts from the bottom of their hearts, and at the same time they are full of pride in what they have done, forming a utopian environment.
04 Technology and Civilization
Writing code on a computer is a dialogue between man and machine; creating a system to realize automatic driving is not only a dialogue between man and machine, but also a dialogue between machines and machines, and a dialogue between machines and civilization. In this more challenging proposition, it also focuses on how autonomous driving can change people’s lives.
In Lou Tiancheng’s eyes, autonomous driving is a game between technology and civilization. And in this game, just like people, technology is a living individual who can think independently; in other words, he gives civilization to technology instead of technology to civilization.
This was first reflected in his choice to switch from Google to Baidu for autonomous driving, and then reflected in the development path of Pony.ai after starting his business.
Lou Tiancheng told Leifeng.com that he joined Baidu after leaving Google because he wanted to do autonomous driving in China. When Google was researching autonomous driving, he found that Google’s understanding of Chinese traffic culture was too far away, and it didn’t understand Chinese driving habits at all. Americans and Chinese have different ways of thinking and driving. Americans are full of rules, but the most important thing in driving in China is communication. (Cars) must understand people’s intentions, analyze and communicate, and form compromises in conflicts with rules.
Lou Tiancheng also emphasized that cars integrate into people’s reverse thinking: “In many cases, what can be done by people, why can’t automatic driving be done? For example, cars don’t understand people’s eyes.” He believes that these are the most fundamental problem, and the safety of autonomous driving is to be like a human being, allowing the car to integrate into the traffic flow and become a part of society.
On the surface, the meaning of autonomous driving is to liberate drivers, but in Lou Tiancheng’s view, the deeper meaning should be that after successful scale, autonomous driving may change people’s car habits, the relationship between people and cars, and the relationship between cars and roads , Car management and so on. Just like computers, computers first appeared to help with calculations, but then many new needs and experiences broke out and changed our lives. Lou Tiancheng believes that the same is true for autonomous driving.
This is also the inspiration Yao Qizhi gave him. Yao Qizhi believes that all technology should be centered on people, and the existence of technology is to change the way people get along with society. The emergence of autonomous driving may stimulate many new needs, and care for the no man’s land neglected by modern civilization (such as desert traffic, travel for the blind, etc.), this is the real “technology leads life”.
It is also in this possibility that Lou Tiancheng finds greater pleasure. A number of people close to Lou Tiancheng told Leifeng.com that they thought they were already very diligent in the research of autonomous driving technology, but Lou Tiancheng is even more diligent than them. If there is a problem, it seems that he can be found 24 hours a day, and the message will be returned in seconds. .
In the matter of making trucks, Lou Tiancheng said that trucks also belong to L4. On this point, he agrees with Peng Jun. They all believe that the automatic driving of trucks and cars is technically similar. Ma Zhixing has been laying out the truck business line since the middle of 2018.
Treating technology as an independent life is reflected in the fact that Pony.ai’s research and development is often “one step ahead”. Just as the growth trajectory of a person can be predicted, the technology research and development of autonomous driving also has its own inherent development rules. Lou Tiancheng can often see in advance what kind of problems need to be solved in the next step of the “growth” of the technology, and then invest in advance.
Such as data storage and processing. Autonomous driving has successively gone through the stages of car modification, data, perception, decision-making planning, system, etc. Now it is focusing on cost, mass production, and safety. Many companies didn’t know about the data age at first, and cut directly from the system and the scene, while Lou Tiancheng came from the data age (around 2013), knew the importance of data storage and processing, and laid out a solution to the surge in data in advance. Li Hengyu remembers that when they only had one or two cars, they only needed a server to store data in the office, but as the data surged, they needed to build a data center, access the cloud, and synchronize data. But these situations all appeared in their planning ahead of time, so there was no confusion.
For example, establish an evaluation system for long-tail scenarios in advance. Autonomous driving needs to achieve 99.99% to be implemented on a large scale. When you want to achieve such high reliability and performance, you have to deal with more and more long-tail scenarios, and these cases are difficult to obtain and the number is very small. . In this case, how to deal with the long-tail scene while ensuring that the overall situation moves forward is a difficult problem. Wang Haojun told Leifeng.com that Lou Tiancheng had foreseen such problems in the future two or three years ago, and formulated a set of evaluation system to ensure that they could finish the last few kilometers of the marathon.
Lou Tiancheng believes that the difficulty of autonomous driving lies in the need to “accumulate”, and there is no magic martial arts cheat that can be achieved overnight. When he looks at Pony.ai’s unmanned car data, he also looks at the data of all the cars added up for several months, rather than looking at the evaluation of one or two days.
But he also believes that self-driving cars can be built in a very short period of time, not 30 or 50 years as many people say. Autonomous driving is something that no one has ever succeeded in. There is no precedent to follow, and they can’t tell how long it will take to reach the destination, but they care more about whether the direction is correct than the length of time. And they’re sure they’re headed in the right direction.
Lou Tiancheng looked at the distance with one eye, and looked at the feet with the other eye.
Both Li Hengyu and Wang Haojun said: Lou Tiancheng is the same as Peng Jun. When playing Texas Hold’em, if the probability of winning is not high, he has a strong ability to give up. They dare to fight hard, but they also have an extremely rational side to risks and benefits.
When choosing automatic driving, Lou Tiancheng already knew that this was a marathon. Entrepreneurship is different from competition. Entrepreneurship must consider commercialization, and lofty ideals also need realistic support. Many people told Leifeng.com that they thought that Lou Tiancheng was a person who could separate idealism from commercialization; Lou Tiancheng’s idealism was not “no money”. Will refuse bonuses, and he will do the same after starting a business. For example, while pursuing the ultimate L4, he also promotes better commercialization solutions.
“Essentially, he is not only an idealist, but also a practicalist.” Wu Chenye said.
05 idealist
In 2005, Tsinghua’s Department of Computer Science introduced the concept of “theoretical computing” into the education system for the first time. The hero behind this was Yao Qizhi, Lou Tiancheng’s mentor and winner of the 2000 Turing Award (the highest honor in the world’s computer field).
Yao Qizhi returned to China, set up Yao Class in Tsinghua University, and opened two courses called “Theoretical Computer” and “Theoretical Computer”. Lou Tiancheng, Zhang Ning, Bei Xiaohui, Ge Rong and others are the first batch of students of these two courses. Lou Tiancheng told Leifeng.com that they had to register for these two courses at the time because Yao Qizhi said how difficult the two courses were when he was preaching, and the word “difficult” aroused their determination to win.
In general, including Lou Tiancheng, this is a group of people who are very proud of their hearts, and many of them have already decided to recommend to Qingbei before entering high school. However, the content of these two courses stumped them successfully. When they first entered the university, they conveyed to them an awe-inspiring cognition of knowledge and technology: learning is endless.
Lou Tiancheng has been acting according to his heart since then, from programming competitions, to doctoral research on data mining, to doing autonomous driving. After graduating from undergraduate in 2008, Lou Tiancheng stayed with Yao Qizhi to study for Ph.D. He was the only one among the ten doctoral students who chose to study data mining. The others were doing the same direction as Mr. Yao, such as theoretical computing and quantum computing. But since Tsinghua University, he has established a high pursuit of technology, and like many students in the Yao class, he aims to expand and make breakthroughs in the boundaries of knowledge and technology like Yao Qizhi.
Autonomous driving is a new field. Just as many ordinary people cannot understand other than geeks, they cannot define: what kind of self-driving company is a successful self-driving company?
如同近十年来迸发的许许多多的新科技产业一样，市场看待他们的目光，无异于看土老板开火锅连锁店，以「打开门做生意」的思维来评价这些需要技术积累与基石的新赛道。相形之下，知识分子是一个外来怪物，理想主义是一个贬义词；星空是遥远的，他们更呼吁「接地气」。
但这两者其实并不相斥。在楼天城的眼里，自动驾驶是技术与文明的博弈，但文明始终是技术发展的核心。小马智行在发展自动驾驶的过程中，也坚持这样的理念。
自动驾驶也是一个难的领域。而一个有能力的人愿意将个人的理想与社会的理想相结合，把少数人做不到的事情担在自己的肩膀上，哪怕只是为一种造福他人的、未经验证的可能性，其实也是时代的幸运。
在谈编程竞赛、谈自动驾驶时，楼天城始终维持一种冷静的态度坐在沙发上。对他来说，他目前在自动驾驶上取得的成绩、甚至过去在竞赛上的排名与荣耀，似乎都只是人生道路上的途经点，不足过分夸耀。他谈起IOI 2004，说「雅典真美，有真正的独到之处」；谈起竞赛，调侃自己在比赛中做过的错题数大概能排第一名，「我曾经经历过四场比赛垫底、单场错题数高达91道」；而自动驾驶，「要积累」。
成功一笑而过，而失败的惨痛更能让楼天城记忆深刻，多年无法释怀。楼天城是这样的一个人。与其说他的人生是在跟竞赛、跟人、跟商业世界对抗，不如说他是在跟「技术」本身博弈。那些失败的回忆里，知识再一次让他胆怯，鞭策他还要不断奔跑向前。
个人或团队的成就，无论多大，往往也只是一个时代下的小插曲。到自动驾驶做成的那一天，相比人们说「了不起的楼天城」、「了不起的小马智行」，或许说「了不起的技术者」更让楼天城热泪盈眶。那一天，楼天城只会说一句：
「自动驾驶做成，我也有所贡献。」
楼天城（左）与雷峰网创始人林军（右）在硅谷
人往往在大的时代中完成个人命运般的选择与迎接戏剧般的结果。但我们仍相信，哪怕在消沉的集体意志中，仍有值得记载的创造激情与放手一搏。长期关注科技人物与他们的故事，如果你有想看的人物与推荐的故事，欢迎添加作者微信Fiona190913 交流。
本文转自：https://www.leiphone.com/category/ai/GumWH5ufxYyBnrpO.html
This site is only for collection, and the copyright belongs to the original author.