Intelligent manufacturing, a industrial competition that cannot be lost | Jiazi Guangnian

A difficult and correct road.

Author | Xiao Nan Xianhuan Edit | Chestnut

When you get a new phone, open the box to tear off the protective film, a brand new, perfect screen appears in front of you. And a cover glass from being produced to embedded mobile phone screen, it takes all its "luck".

Covering glass is also called window protection glass in the industry. It is a transparent lens used to protect the touch module and display module. As the upper protective layer of the screen display and the touch part, it is the most frequent hardware in contact with humans on mobile phones, even one of them.

Not every cover glass is perfect.

Covering glass production involves multiple processes such as opening, CNC carving, grinding and polishing, cleaning, tempered treatment, silk printing, baking, and other processes. Throughout the process, tens of thousands of glass will go through multiple detection links. At least a few years ago, this session was still carried out by manual, ensuring that each cover of the glass was perfect when the factory was exported, just like the finished product you finally saw.

This process is far from being so relaxed.

The defects of the cover glass can be divided into more than a hundred types. The quality inspection session covers corners, white glass detection, and silk printing. On a production line, the cover glass that needs to be tested every month reaches hundreds of thousands of pieces.

The complexity of the detection scene is actually a microcosm of my country's manufacturing industry. The technology has not yet reached this trillion -scale industry, and various aspects such as design, production, management, and services are full of too many variables and unknown.

In the view of He Zhiqiang, president of Lenovo Venture Capital, opportunities are hidden in it.

"In the Internet era, we have seen core technologies such as AI and big data, allowing giant companies to continue to understand the deep laws behind the data, quickly iterate the products, and continuously improve their competitiveness. For any industries, manufacturing is the most important main track. To open the barriers of software and hardware, industrial data is constantly accumulating and iterative, and the manufacturing industry will be subverted. "

When data, robots, and AI enter the main track, the industry may be changed, even if it is just a small mobile phone cover glass detection ...

And also looking for technology to reshape industrial opportunities, there is a group of scientists who came out of the laboratory and scientific research institutes.

Occasionally, Zhang Zhengtao, the chairman of the founder of Zhongke Huiyuan, walked into a glass testing plant in Suzhou. What he saw was a dim, forced dark room, and a group of young girls and young men stood on the workshop to detect glass defects under the light of 12,000 Lax.

You know, the brightness of the human eye is about 500-750 Lax. 12000 Lax means that this is dozens of times the extreme working environment of normal brightness. After almost 3 months of quality inspection staff, vision will drop to 0.6.

The moment he really stepped into the factory, Zhang Zhengtao was shocked by the bright and dark scene in front of him. The use of a bottle of eye drops of workers is like drinking water, and the pain is constantly swelling.

"I didn't expect to have such a high -harmful job. At that time, I felt that I would solve this problem."

However, correct and magnificent significance, everything is perfectly in line with what you want. At that time, the detection technology of mobile phone glass covers was monopolized by Germany for a long time, and companies in many countries around the world had hit the wall.

In the face of such a difficult mountain, "fierce war" is inevitable.

1. The existing nails, the hammer to be polished

In the long link chain of glass production, "the world is bitter for a long time."

Relevant statistics show that more than 3.5 million workers in China conduct product appearance inspections on the production line every day, and more than 1.5 million people in the 3C industry.

Workers are scarce resources.

The pain points are already available, how to polish the "hammer" that solves the "nail".

Machine vision provides a new imagination space.

At that time, the two teams did not meet at the time, but at the same time, they saw the opportunity of the machine vision based on 5G+AI technology, which will bring the opportunity to improve the efficiency of the detection link.

Zhang Zhengtao is an atypical entrepreneur. Prior to this, the Laboratory of the Chinese Academy of Sciences was his main position and focused on scientific research. I don't know much about the industry. What is the field of mobile phone cover glass detection, what links in the industry, and what difficulties, he does not understand.

The decision made by the head of the head, but the specific and clear difficulties came from the welcoming.

In the first year of the establishment of Zhongke Huiyuan, Zhang Zhengtao has been busy solving the problem. He understands computer vision technology, but does not know how to design the optical system of highly transparent, high -reflective defective and flawed imaging such as cover glass? What is the craftsmanship of the cover glass? People who understand the algorithm in the team do not understand the craftsmanship, and those who understand the crafts in the factory do not understand the algorithm. The gap between the two sides is deeper than that of the East Africa Great Rift Valley. How to solve the problem if you don't enter the scene?

"Basically, 70 % of mobile phones in China are made in China, but from the perspective of factory or workers' labor protection, technology uses technology to assist artificial quality inspectors to solve this problem." Song Chunyu, a senior partner of Lenovo Venture Capital, said to Jia Ziguang.

This also explains why Zhongke Huiyuan was not designed in 2017, and Lenovo Venture Capital invested them. Zhongke Huiyuan has almost learned about glass cutting, tempered, silk printing and other processes from scratch, and is familiar with and understands workers' needs over and over again. This is a long process. It took almost a year to find out the process. The design of the corresponding algorithms, optical systems, and prototype tests took more than half a year. The first -generation finished product test instrument did not gradually take shape until 2018. Fortunately, the result is good.

The cover glass finished detector developed by Zhongke Huiyuan can complete the detection of a mobile phone cover glass in 1.5 seconds. Compared with artificially, the efficiency has increased by more than 20 times.

One day in April 2019, Huizhou, on the production line of a large global large glass cover board manufacturer, was established for nearly 3 years. Place it, and perform the appearance defect detection and verification of the glass cover. The surrounding people are full of people, including the core leaders of both parties. Zhang Zhengtao's heart was a little embarrassed.

After the machine was running normally, a hanging heart finally let go.

The results show that the indicators of Zhongke Huiyuan's testing equipment are better than overseas equipment. And for the scope of detection, Zhongke Huiyuan can achieve full area coverage. In this regard, the first self -developed AOI testing equipment in China has officially passed the verification, and the monopoly situation of overseas equipment is about to be broken.

"Chinese equipment is not worse than foreign." The recognition and batches from the industry benchmark customers have made Zhongke Huiyuan's exploration and research and development on the road of AOI detection equipment. More than ten self -developed AOI testing equipment came out one after another, and more and more customers came to the door ...

For manufacturing with relatively low degree of digitalization and experience first, transformation is a difficult process. But once it is completed, an industrial opportunity of N times will be born.

"Data intelligence will reshape the manufacturing industry. This is our Believing. Among them, there will be a new species that will be closely integrated with the industry." He Zhiqiang told Jiazi Guangnian, "Since the establishment of Lenovo Venture Capital in 2016, it has begun to surround intelligence. There are two main points in the manufacturing layout: first, the needs of the industry, the most important demand for real and urgent; the other is whether technology can solve the problem, to what extent can it be solved? Incorporate? "

As long as these two points are established, they have huge development potential. "The manufacturing scene is large, broad and world, and there is something big."

At present, the layout of Lenovo Venture Capital in the field of machine vision has become a system, including China Science and Technology Huiyuan, Simico Technology, Bai Mai, Shenhui Vision, etc., involving many industries in the manufacturing industry.

Established in 2019, Siman Technology, founder and chairman, is a lifelong professor at the Chinese University of Hong Kong. In 2020, Lenovo Venture Capital noticed this startup.

At the beginning of the establishment of Silima Technology, the semiconductor industry was selected as the first stop of the technology landing to overcome the problem of testing the defect detection of semiconductor products. As a high -end precision product, each process step rate in the process of semiconductor manufacturing must be close to 100%to ensure that the final production yield is maintained at the acceptable level. Therefore, the difficulty of detection of semiconductor products can be seen.

Siman Technology is based on its own deep technical precipitation. In just more than a month, this "difficult bone" is used to solve the problem that plagues the manufacturing industry for many years in an intelligent method.

In the most dense and rich Chinese manufacturing industry in the industry chain, the "nails" of demand are spreading all over the chain, waiting for the "smart" hammer.

2. Data change experience, difficult mountains

Before entering the glass testing factory in Suzhou, Zhang Zhengtao had conducted several months of investigations, with a variety of directions, even includes mustard packaging machines. But just like the other dozens of other directions he abandoned, the mustard packaging machine cannot be done or not, but that what I want to do in my heart has always been the most urgent demand in the industry: even "the highest industry industry is the highest in the industry industry, the highest industry industry is the highest in the industry. "Mount Everest" also crossed it.

But in a huge and complex manufacturing system, the mountains that are difficult to overcome, more than this?

One of them is industrial software.

Industrial software is accompanied by industrial development, and the two complement each other. The industrial software has a high threshold, which requires the simultaneous software development skills and rich mathematics, physics and industrial knowledge. The R & D cycle is long, the iterative speed is slow, and the customer uses high viscosity. According to Jiazi Think Tank's "2021 China Science and Technology Investment Report", the domestic penetration rate of China's industrial software is only 6%, which is the "card neck" industry.

In 2016, compared with the excitement of the consumer Internet, industrial software seemed very unpopular.

"Industrial software is a large number of industrial practice experience and the condensation of industrial hidden knowledge. It is indispensable for the transformation and upgrading of China's manufacturing industry, and it is of great strategic significance." He Zhiqiang believes that from the perspective of the long -term pattern, there must be more in China's industrial development to today. Core technology should be in your own hands. The design and development of our own industrial software is urgent.

In 2017, Lenovo Innovation Investment was decisively investing in Nanjing Tianyu, which was not famous at the time.

The latter has been established as early as 2011. It focuses on the development of industrial software with independent intellectual property rights, integrates AI technology with CAE technology, and has developed a smart design software, intelligent simulation analysis software, intelligent optimization software, and intelligent data construction. Full -chain software products covering industrial products such as molding parts. He Zhiqiang, who has a profound understanding of technology, saw the opportunity to be hidden inside early. He believes that my country will mobilize more and more forces to develop industrial software in the future. Although there were not many customers at that time, Tianyi customers were not as many as large foreign manufacturers, but this was a high -threshold area with technical barriers, and industrial software was used. The more and better, the more feedback from. The better, the development space is very large.

In this way, Nanjing Tianzhang became the first attempt for Lenovo Venture Capital in the field of intelligent manufacturing. It turns out that this is the right choice.

"For high -end industries, the performance and safety of the product are critical. Even if it can only improve the performance by 1%, many companies are willing to spend money to do such a thing." year.

For example, a car manufacturer will not make more than a dozen cars hitting each other for testing performance. The cost of doing so is high. No company will do this, let alone enterprises in the fields of aerospace, ships and other fields. Therefore, Tianzheng currently focuses on medium and large enterprises.

If you are a ship manufacturer, you need to put it in the water to experiment after the ship is made, but if you find that there are problems in the experimental stage, things will become tricky. Essence The role of industrial simulation software is here. Song Chunyu Jiazi Guangnian can simulate simulation and simulation on the computer during the design stage of the ship. View its speed and the performance is inaccurate.

"At that time, it took more than 20 days to simulate a boat under normal circumstances, and Nanjing Tianyu could be completed in a few hours, which greatly improved the design efficiency. The huge improvement of this efficiency was what we were looking for." Song Chunyu was looking for it. "Song Chunyu. Said, "Of course, we also have to see if this product has more advantages and prospects in China, and matches the industrial chain high."

With the past two years, the "domestic replacement" of industrial software has become higher. In the market, in He Zhiqiang's view, the investment of Lenovo Venture Capital is not so much investment in "domestic alternatives", and it is more like the logic of in turn. "Domestic alternatives are just a bonus."

Whether it really meets the needs of the industry is the core.

These industrial software companies that are also bred have the potential to become future giants.

3. The machine enters the factory, faster, faster

Some people climb the peak of software, and some people climb the peak of hardware. In the industrial manufacturing industry, there are also robots that cannot be avoided.

Robotics are not only highly hoped in the quality testing link. As the transformation of traditional industries has accelerated, it has become an important trend.

In the field of logistics, companies often count such accounts: buying an ordinary artificial forklift only cost tens of thousands of dollars, and it takes hundreds of thousands of dollars to buy a unmanned forklift. If it is you, how do you choose? Buying the most cost -effective items at the least price is the "nature" of businessmen. If this is a simple arithmetic question, the answer must be the former, but is this a simple arithmetic question?

According to GGII data, from 2016 to 2021, the sales of unmanned forklifts in the Chinese market increased from 350 to 7,375 units, with an average annual compound growth rate of nearly 66%. It is expected that the market growth rate of about 50%in 2022 is expected to maintain a market growth of about 50%. Essence

Why do so many people choose "wrong"? Then come to calculate another account: artificial forklifts need to be manual, high -level forklift type drivers need to hold a certificate and have a long training cycle. If the "old man" after training is left, the "newcomer" will go through another round. Cultivation, this does not include the salary of workers who need to pay. A more severe problem is that even if it is "high salary", it may not be recruited.

The founder of a robotic company revealed to Jiaziguang that in the process of visiting customers, they found that a chip sealing and testing manufacturer's monthly worker resignation rate reached 25%, which was equivalent to a wave of people in the first quarter.

Lenovo Chuangjun partner Wang Guangxi was also shocked by the scenes of the "mobile army" outside the factory and the operation of the assembly line in the factory. "Many production line workers cannot sit, because they will be stunned, distinguished, and even dangerous. Standing for a few hours of assembly line operations, it is very hard." China is a large industrial country and a manufacturing country, and will continue to maintain the status of the world factories. Then Labor has become a problem that must be solved.

In fact, Lenovo Venture Capital began the layout of the industrial robot field in 2017. On the eve of the outbreak of robotics, a comprehensive layout was made around the manufacturing of upstream parts, manufacturing in midstream, and integrated downstream integrated applications. Under the double catalysis of the epidemic and national policies, industrial robots quickly entered a large number of new industrial scenarios in 2022.

"Every year we discuss the critical point of the explosion of robots in each process. At the same time, we will also think about technical difficulties, such as collaborative robots. There are only tens of thousands of shipments a year, which cannot be in the industry. We studied how to solve this problem with people in the industry. "Wang Guangxi told Jiazi Guangnian.

In the future, robots are a typical representative in the field of industrial robots.

Future robots were established in 2016 and entered the industrial unmanned vehicle track early. This is also a startup company that came out of colleges and universities. Behind it is Professor Liu Yunhui, director of the Institute of Tianshi Robotics, the Chinese University of Hong Kong. In the future, the investment of robots is not so "silky".

In 2018, Wang Guangxi met Li Luyang, co -founder of the future robot, but the real investment was in 2020. The turning point of the story is that Wang Guangxi feels that Li Luyang has a powerful "self -iterative upgrade" ability.

When the two just met, Li Luyang was still a doctoral temperament. He was good at learning new things. He was young and strong, but after two years of knowing, Li Luyang successfully "upgraded". On the one hand, it looks old and skilled. The timing of investment has arrived. "His changes may drive the level of transformation that others may not be able to achieve in 5 years or even 8 years in a year." Wang Guangxi told Jiaziguangnan that the quality and growth potential of the founder is also an important for Lenovo Venture Capital to consider being invested by the investment enterprise. factor.

In addition, more importantly, compared to manufacturers that provide solutions based on lidar -based technology, the solutions provided by robots in the future are based on computer vision technology. The cost is lower and can recognize color. It has unique advantages in many scenes that need color recognition.

Based on the full -stack solution of the top -level system+terminal industrial unmanned vehicle, the unmanned forklift provided by the robot in the future can flexibly "walk freely" in the storage environment, including high and low warehouses, from 5, 6, or even 9 meters. The place is flexibly pickled, put on goods, stack, and is efficient and convenient.

"Under normal circumstances, 4 unmanned forklifts can replace 8 and 3 artificial forklifts." In the future, Xie Li, director of robotic marketing, told Jiaziguangnan that if the company was initially worried that they could not recruit employees, then as unmanned forklifts The advantage appeared, and the factory's unmannedization gradually became the strategic layout of the enterprise, so it would cause blowouts in the entire industry.

The entire industry was still early, and the blue ocean was emerging.

4. Long road resistance, intelligent manufacturing span deep water zone

To this day, Zhang Zhengtao still remembers how difficult the first customer was found.

At that time, domestic quality inspection was blank. Except for Zhongke Huiyuan, there were no other companies doing it. How to find customers and obtained their products to their online practice. The entire team's mind was also "blank."

Occasionally, Zhang Zhengtao learned that there was a glass cover board manufacturer in Songshan Lake in Dongguan, Ruilidida, who was undertaking the Industry 4.0 project of the Ministry of Industry and Information Technology to try to solve the problem of quality inspection with technology. As a result, Zhang Zhengtao hurriedly took a wave of people to Songshan Lake, rent a house, recruit soldiers to buy horses, and asked business colleagues to squat at the door of the factory.

In fact, there are more than just Zhongke Huiyuan.

In November 2016, Nanjing Tianyi received an investment of 15 million yuan in Lenovo Venture Capital. In the next four years, no investment institutions have voted for it until 2020. Immediately afterwards, Nanjing Tianyi continued to honor the honor, and its voiced reputation in the industry ushered in highlights.

The reason why we ushered in gorgeous turns is not only because it is excellent, but the deeper reason is that people are increasingly aware of the general trend of intelligence, and the entire hard technology track is getting hot.

As an earlier entry of Lenovo Venture Capital, it has reinstated the logic of the industry and the data to reshape the industry, and has invested a number of industry leading enterprises in the field of intelligent manufacturing. Many companies they have invested in have cooperated to piece together a complete intelligent manufacturing landscape.

He Zhiqiang believes that the reason why it is so early is because Lenovo Venture Capital is a CVC, and the industry is in front of the industry. It has the industrial resources and manufacturing field Know-How, forming a unique investment idea and resource advantage. "At that time, when we started to vote, this was not a popular track. But we always believe that the smart Internet, or the industrial Internet now talks about the change of industries in the next ten years and even longer cycles, ","

With the continuous upgrading of intelligent manufacturing, Lenovo Venture Capital also focuses on a farther future. "We have invested in the foundRial Next to help the OEMs create an independent factory this year. It has been verified on the TESLA production line. It has helped enterprises to gradually achieve intelligent production and autonomy of the factory through intelligent technology, and finally realize software definition manufacturing." Song Chunyu Tell Kaizhi Guangnian.

The transformation of fragmented industrial scenarios is an important direction, but the core competition of industrial manufacturing scenarios in the future will return to the optimization and upgrade of research and development and manufacturing capabilities. In all aspects of production, data to achieve closed loop in various scenarios will make the manufacturing plant continuously iterate and achieve traditional disruptive development, which will be the ultimate goal of intelligent manufacturing.

"We are not limited to domestic, but we must be able to be more competitive in the international market." He Zhiqiang believes that the ultimate goal of Lenovo Venture Capital is to help my country go from a manufacturing country to a manufacturing power. "This is a difficult and correct road."

- END -