Author Han Xiao

Editor Han Xiao

Another autopilot company has got financing!

Recently, Ling Jun Science and Technology of L4 Robotaxi Company announced that it has obtained a new round of financing of tens of millions of yuan. According to the industrial and commercial data of Tianyancha, the investment was led by the freight industry "Didi"-the American listed company Manbang Group, and the old shareholder Zhenxin Capital followed suit.

Although Ling Jun Science and Technology was not well-known before, it can be seen from the recent autopilot financing news including this one that all kinds of investors are increasingly favoring potential companies with late-comer advantages.

This phenomenon is not difficult to understand. After several years of development, head players face three major dilemmas after reaching high valuation: the listing path of US stocks is not smooth; There is no obvious gap with the rear players in technology, and the realization of driverless driving is still far away; Insufficient hematopoietic capacity, the company is always facing the risk of food shortage …

In short, the industry has entered a long and painful marathon stage, and investors from all walks of life naturally pay more attention to those latecomers who seem to run slowly but have good team and technical ability.

With the founder of Ling Jun Science and Technology &; In the words of CEO Yang Wenli, "We spent tens of millions to realize Robotaxi that others spent billions."

Yang Wenli, CEO of Ling Jun Technology

Although his words are not long, his confidence in his own technology is revealed between the lines, and it also shows that Ling Jun Science and Technology has a different understanding and style of self-driving entrepreneurship. After several hours of communication with Pang and Pang of Science and Technology, Che Dongxi was able to show you more information about the old gun entrepreneurial team of Science and Technology, and deeply interpret this "slow" development model.

Ling Jun Science and Technology was founded on 20 16. In terms of time, it belongs to the first batch of startups in the industry, but it has been extremely low-key and even marginalized in the past six years.

"On the surface, we have less money and a small team. In fact, we deliberately slowed down the development speed." When it comes to development, Yang Wenli explained to Che Dongxi, "We are actively controlling the speed of development."

Leading technology self-driving test vehicle

This is directly related to the judgment of Yang Wenli and the founding team on the development of the industry. In their view, the development of the autonomous driving industry is divided into four periods:

1, core supply chain maturity

The time range is from 20 13 to 2020. At this stage, the core supply chain of autonomous driving is immature and expensive. The main job of each company is to polish technology and try mass production.

2. Landing date of small-scale commercialization

The time range is 202 1~2025. At present, with the increase of demonstration cities, unmanned minibuses, intelligent networked buses, feeder unmanned logistics vehicles and other technical products can be applied on a small scale, and enterprises can earn a certain income.

3. Market scale expansion period

The time range is from 2026 to 2030. At this stage, with the maturity of technology and the further opening of regulatory laws and regulations, autonomous driving technology has begun to land in more and more scenes and scopes, and the industry pattern has begun to be reshaped.

4. International expansion period

The time range is after 2030. This stage is basically stable with the end of domestic competitions. Players need to explore the international market for further development, and at the same time need to operate a large-scale technology company in a refined way.

This kind of industry cognition has guided the development thought and rhythm of science and technology in Ling Jun.

For example, in the first stage, the industry generally faced the problems of lack of wire control chassis and poor performance of lidar. This time is obviously not suitable for large-scale expansion. If a lot of people are recruited to make great efforts to solve these parts problems, by the second stage, you will find that the supplier has solved all the above problems, which is equivalent to doing nothing before.

Another example is the second stage, which is not suitable for large-scale expansion when the technology is not mature enough.

"At this stage, if hundreds or even thousands of test vehicles run all over the country, once the technical structure changes or is greatly upgraded, the existing fleet will become invalid assets and it will be difficult to deal with." Yang Wenli said.

There are obvious cases of this phenomenon at home and abroad. For example, this year Waymo released the fifth generation Robotaxi, which has been upgraded in terms of body, sensor and computing platform, while the old models cannot be directly upgraded to the fifth generation, and the fleet needs to be gradually replaced.

In addition, the autonomous driving industry has just entered the second stage, and the overall revenue capacity is not strong, relying on financing blood transfusion.

"Large-scale, high valuation, future financing will become more and more difficult, and finally have to lay off employees or go bankrupt." Yang Wenli analyzed the car. "But the small and beautiful team can break even in the second stage and guarantee to live to the next stage."

One day on 20 15, a BMW 3 Series GT painted in orange and white with a "big flowerpot" on the roof drove to Beijing G7 Expressway, then turned into the Fifth Ring Road and turned a corner back to Baidu headquarters in Xi 'erqi.

This BMW is the starting point of domestic commercial autonomous driving research and development. It was built by a team of more than 20 people from Baidu and BMW. In the past five or six years, more than 20 people have left Baidu and established a number of autonomous driving companies, which has propped up half of the domestic autonomous driving industry.

Yang Wenli is one of these 20 people, and he is an old self-driving gun.

He completed his undergraduate and master's studies in the Department of Automation of Tsinghua University, obtained his doctorate from Pennsylvania State University, and worked as the chief architect of Western Digital.

After returning to China, Yang Wenli joined Baidu's cutting-edge Deep Learning Institute at that time and participated in the establishment of Baidu's early autonomous driving team.

In 20 16, Yang Wenli and his Tsinghua alumni Han Yan and He Jiarui founded Ling Jun Science and Technology, and the research and development goal was locked in the most difficult L4 Robotaxi direction.

Han Yan holds a Ph.D. from Tsinghua Automation Department and is the senior vice president of map and simulation research and development of Ling Jun Science and Technology Company. He is a successful entrepreneur. The first startup he participated in after graduation has been listed. Before joining Ling Jun Science and Technology, the second joint venture company was acquired by Qihoo 360 and became a senior R&D engineer of AI, responsible for the research and development of basic technologies in the fields of operating system simulator, network security and pattern recognition.

He Jiarui, Master of Automation Department, Tsinghua University. He used to be a senior R&D engineer in Baidu's Autopilot Division, and now he is the senior vice president of R&D for decision-making and planning in Ling Jun Science and Technology.

He holds a Ph.D. degree from the University of Tokyo, and the former founding partner and CTO of Lu Ling Technology, Si Ruochen, is the senior vice president of R&D of technology perception system in Ling Jun.

In addition, Ling Jun Technology also has two partners, Pang, who are the Chief Financial Officer and Pang. Deng Haiqing is a postdoctoral fellow of the People's Bank of China, an independent director of Ganzhou Bank, a visiting professor at the National People's Congress and Beijing Normal University, and has rich financial experience.

Pang Dongjun, Chief Operating Officer of Ling Jun Technology

Pang Dongjun has worked in the autonomous driving industry for many years. He was in charge of business and key customer business in Tagg Star and E-control House, and then participated in the establishment of business vice president in Wan Zhi, with rich experience in technology application.

As can be seen from the background of the core members, Ling Jun Science and Technology not only has strong technical capabilities, but also pays attention to introducing professionals in finance and technology landing, and has built a perfect team with complementary advantages.

High-quality teams are naturally sought after by employers. 20 17 and 202 1 have successively received angel round investment from Wu Yuefeng, Jiuhe and Xintian and Pre-A round financing from Zhenxin Capital. In addition, domestic well-known AI chip companies such as Horizon and Ganzhou Development Investment are also investors.

In addition, it should be noted that the six years since the establishment of Ling Jun Science and Technology have been the most lively time in the global autonomous driving industry-entrepreneurial waves have come and gone, and a lot of financing has continued. In the past more than 2,000 days, various peers, automobile companies and travel companies have been throwing olive branches to Lingjun Technology or team members-hoping to acquire or poach its technical experts.

However, its core team is completely unmoved by short-term interests and has a firm development direction. So far, no one has left, which is a clear industry judgment and tough strategic determination at work.

In terms of technology research and development, the development strategy of Ling Jun science and technology is "pearl mining, dimension reduction and commercialization".

In the eyes of Yang Wenli and the leading technical team, urban scene driverless technology (Robotaxi) is the crown jewel of the industry, so the company has locked in this direction from the first day, hoping to finally win this pearl.

However, because Robotaxi technology is the most difficult and takes a long time to realize, it is necessary to apply technology to reduce dimensions in specific scenarios during the research and development process. One is to earn money to support the team through commercialization, and the other is to collect data to promote Robotaxi technology iteration.

Adhering to this development idea, Ling Jun Science and Technology chose Robobus self-driving bus and urban feeder logistics for dimensionality reduction.

The driving scenes of buses and feeder logistics vehicles are exactly the same as Robotaxi, but the driving speed is relatively slow, and most of them are fixed routes, which is very suitable for building products based on Robotaxi technology and landing commercial use.

Pang of Ling Jun Science and Technology told Chedong that they had built two Robobus, one for the unmanned minibus in the park and the other for the intelligent networked bus, with a length of 5.9 meters. Both vehicles have been put into normal operation in the New Energy Automobile Science and Technology City of Ganzhou Economic Development Zone, Jiangxi Province.

Ling Jun technology's self-driving minibus

Driverless minibus is a low-speed vehicle without steering wheel and operating pedal, which mainly runs in semi-closed scenes. The intelligent networked bus runs along a fixed route between the airport, the railway station and the Economic Development Zone Management Committee.

During the operation period, the unmanned minibus runs every three days at the longest, and no one takes over. The working hours of the daily shift are 8 hours, and the operating mileage is about 150 km. In other words, it has traveled 450 kilometers in 24 hours in urban scenes without any takeover, which shows that its autonomous driving technology has reached a high maturity.

It is worth mentioning that Ling Jun Science and Technology adopts the collectivization development strategy. In the landing area, in addition to the driverless minibus and the intelligent networked bus, its Robotaxi will also come for testing. The three models test and collect data at the same time, thus promoting technical iteration and ultimately improving the driving performance of the three models at the same time.

Ling Jun technology self-driving CMB

Since last June 1 1, the three test vehicles have been put into normal trial operation for five months.

Robotaxi model of Ling Jun Science and Technology also has excellent performance. According to Yang Wenli, although its Robotaxi fleet is small, it already has the ability to realize P2P (from parking lot to parking lot) in the city, and can cope with all driving scenarios including parking, ordinary roads, intersections, expressway, roundabout and tunnels.

"Others spent billions to achieve it, and we spent tens of millions to achieve it." When talking about the technical performance of his unmanned vehicle, Yang Wenli gave such an evaluation of this vehicle with a smile.

Robotaxi, or automatic driving in L4 city, the biggest difficulty at present is mainly the game with other traffic participants, that is, the decision-making scale. Ling Jun Science and Technology solved this problem through a number of technological innovations, such as mixed decision model, data rumination and simulation test.

Unmanned vehicles in urban scenes will encounter almost endless special situations, and the rule-based decision-making model can't cope with them at all.

Ling Jun Science and Technology then combined the rules with AI technology, which used deep learning technology to learn how human drivers handled various special situations, and then used rules (such as not pressing the line or running a red light) to ensure that the driving decisions given by AI algorithm were safe and in line with the rules.

This not only gives full play to the advantages that AI technology can solve special problems, but also avoids the black box problem of deep learning model, which has the best of both worlds.

Of course, the premise of the above approach is that there are enough special scene data to promote technical iteration. Ling Jun's fleet of science and technology is not too big. Where does the data come from? How to process data? How to apply data for iteration?

Yang Wenli introduced a closed loop of data research and development.

First of all, its products such as Robobus, which operate normally, collect a large amount of data every day. After getting the data, Ling Jun Science and Technology will analyze the data at the semantic level, extract multiple independent scenes and build an incremental scene library.

Secondly, it will put the scene library into the simulation engine and randomly arrange and combine with the existing scenes to reconstruct hundreds or even thousands of kilometers of virtual test scenes, which greatly improves the efficiency of data application.

Finally, with the help of a larger test scenario, Ling Jun Science and Technology can polish its own algorithm, improve the performance of autonomous driving products, and form a closed loop of data research and development of "product-data-scenario-simulation-algorithm-product".

In addition to the closed loop, Ling Jun's R&D ideas from beginning to end are also quite eye-catching.

In Yang Wenli's view, the ultimate and most important goal of the automatic driving system is to make a good decision-making scheme, so the essence of doing a good job in the automatic driving system is to do a good job in the decision-making scheme system.

In addition, the mass production of software and hardware should be considered at the beginning of research and development technology-developing the system with technologies or hardware that cannot be mass-produced, which will eventually lead to the mass production of the whole system.

Under the guidance of these two principles, Ling Jun Science and Technology first designed the decision-making planning system, then put forward the requirements for the sensing system, and then determined its own sensor hardware configuration according to the requirements.

Therefore, a series of "abnormal operations" appeared in Ling Jun Science and Technology.

For example, because the 360-degree mechanical laser radar is too expensive, it does not conform to the vehicle regulations and affects the vehicle appearance, Ling Jun Science and Technology refuses to use mechanical laser radar on the roof, that is, it abandons the laser SLAM technology and turns to visual SLAM.

Leading technology self-driving test vehicle

In the process of R&D, Ling Jun Science and Technology is the decision-making planning team. If KPI is given to the perception team, the perception team can concentrate limited energy to achieve the most important perception results. In some large autonomous driving companies, the perception team is often in the first place in the R&D process. They provide perception results with their own understanding, and then hand them over to the decision-making planning team for decision-making planning according to the existing perception results.

"Perception has a big problem first. The perceptual results required for decision-making are often not given, but a lot of manpower and computing power are used to identify some unnecessary goals and results, which is inefficient. " Yang Wenli commented on the car.

Two figures are enough to verify the achievements of Ling Jun's scientific and technological practice.

First of all, its autopilot algorithm can be deployed in an embedded controller with only 30 watts of power consumption, which shows that its algorithm is very simple.

Second, there are less than 100 employees in Ling Jun Science and Technology Company. Such a small team can put three kinds of self-driving vehicles into normal operation and run well, which shows that its research and development efficiency is very high.

At the end of the exchange, He Pang also introduced the next mass production plan of science and technology to Che Dong.

In 2022, its Robobus will be put into operation in Suzhou, Hangzhou, Nanjing, Wuhan and other cities, and it is estimated that about 60 sets will be put into operation.

The first batch of self-driving logistics vehicles built in cooperation with Manbang and Shaanxi Logistics will also be put into trial operation. The total number of buses, logistics vehicles and Robotaxi will reach 65,438+000, which will bring tens of millions of operating income to the company.

Ling Jun Technology Beijing Office

After several hours of communication with Yang Wenli, Che Dongxi deeply felt the characteristics of Yang Wenli, which is also the characteristics of Ling Jun Science and Technology.

As a technical expert, Yang Wenli's first impression is a little introverted-her voice is not loud and her speech speed is slow. But when it comes to autonomous driving technology and industrial development, he is like a different person with endless technical terms and opinions. He shrugs from time to time and tells industry jokes, which is enough to show his love for the autonomous driving industry.

The same is true of Ling Jun science and technology. It seems "unattractive" and "unknown", but it has great potential-innovative and pragmatic new ideas are put forward in technology research and development. L4 self-driving products have been mass-produced with little funds and teams, and the development path and rhythm can be strengthened.

To sum up, although Ling Jun Science and Technology was not in the center of the stage before, it is possible to achieve a counterattack in the self-driving marathon.

After bidding farewell to the car, Yang Wenli walked to the office across the street in the darkness.