Location: Home >Posts> You must choose Zhou Jian: It will take ten years for robots to become popular and 90% of startups will die

You must choose Zhou Jian: It will take ten years for robots to become popular and 90% of startups will die

AsiaIndustrial NetNews: Owning a mobile phone of your own and connecting thousands of worlds is no longer a dream.

In the next era, have one of your ownrobotAccompanying entertainment is no longer an imagination for Ubisoft.

“I believe that no matter which IP product, as long as he wants to be a humanoid robot, he will find Ubisoft in the global search. From the perspective of mass production, scale and stability, there is no company in the world that can currently make the same product as us.” Zhou Jian, CEO of UBTECH, said confidently.

Ubisoft Robot, which once became popular in the Spring Festival Evening, has launched several humanoid robot products, and the total shipment volume has exceeded 300,000 units. Since its establishment in 2012, UBTECH has successively launched the humanoid robot Alpha 1, the platform-level intelligent robot Alpha 2 and the Jimu robot for the STEAM education market. The company also launched Lynx, a humanoid robot embedded with Amazon’s voice assistant Alexa, and a commercial service robot, Cruzr, at CES this year, which have formed a relatively complete product line.

The reason why Zhou Jian is so confident actually comes from the real consumer-oriented and mass-produced humanoid robot system he established, the most important of which is the servo steering gear. It is understood that the research and development of UBTECH’s servo steering gear has lasted four years, and currently has a production capacity of tens of millions. At the same time, it is sufficiently commercialized in terms of cost and stability. Behind the servo steering gear, the real advantage of UBTECH is that it is a rare consumer-oriented market in the world, and at this stage it is capable of mass-producing humanoid Robots.

With a valuation of $1 billion in Series B financing, UBTECH quickly entered the ranks of unicorns. At present, UBTECH has nearly 5,000 retail outlets and experience stores around the world, cooperates with giants such as Apple, Amazon and Best Buy, and is building offline flagship stores to enhance user consumption experience.

However, the maturity of humanoid robots is far from easy. In addition to the strong hardware manufacturing and driving technology capabilities, it is the immaturity of artificial intelligence technology.

Sci-fi blockbusters bring us a lot of imagination. Everyone hopes to have a robot of their own, which can move as nimble as a human being, and think and communicate like a human brain. Zhou Jian broke through the first step, but has not broken through the second step.

“”As a start-up company, we face some problems, we have speech recognition (provided by the shareholder iFlytek), machine vision (introduced computer vision expert Dr. Tao Dacheng) and so on. However, a lot of content, a lot of technology (semantic understanding, etc.) cannot be done by UBTECH, which is why we have been financing. ” Zhou Jian said.

But capital blood transfusion alone cannot fundamentally solve the problem. According to Zhou Jian’s theory, UBTECH adopts the method of “COO raising CTO”, supporting itself through sales and investing half of its revenue into technology research and development. Zhou Jian said that UBTECH will use robots as a carrier to build an artificial intelligence technology platform, and launch a robot operating system and a human-like home service robot developed for commercialization within three years.

The upsurge of artificial intelligence has given the public countless expectations, but the reality tells you that this is only the “early stage”.

“I have always believed that China’s artificial intelligence, including robot entrepreneurship, is a myth. I have always believed that 80%, 90% will die.” Zhou Jian said bluntly.

In Zhou Jian’s view, the most important thing for artificial intelligence at present is that it needs to be truly implemented, and it cannot be done just for the sake of the wind. Now many companies are doing pseudo-demand, or products in the transition period.

“In the short term, it will take about ten years for a human-like humanoid robot; in the long run, it will take three to fifty years for a real humanoid robot to be truly commercialized.” Zhou Jian said.

You must choose Zhou Jian: It will take ten years for robots to become popular and 90% of startups will die

The following is based on Zhou Jian’s Q&A record, and NetEase Intelligence has made a deletion that does not change the original intention:

Promote robots into mass production mode, this year’s target sales of 1.5 billion yuan

Q: Alpha is mainly an entertainment companion robot. The recently launched Alpha 2 is priced at nearly 20,000 yuan, which is very different from the first-generation price of 3,999 yuan. Why?

Zhou Jian: The positioning of Alpha2 is different from that of Alpha1. Alpha1 simply has 16 free servo servos for programming, which is similar to toys and education, but the positioning of Alpha2, we hope to have an intelligent The brain, that’s why we took a whole different design direction than the Alpha1.

1 2 3 4 Next > page

As for the price, this is a business strategy of our company, but there will be some adjustments in our price, because Alpha2 is only in China, and all our overseas versions will be changed to Lynx, which is to cooperate with Amazon. But Alpha2, if you know the history of Ubisoft, you know Alpha2 Delete for a long time, but this year is the year when Alpha2 really entered the family.

Alpha robot generation product

Q: UBTECH recently launched the Cruzr robot, which is mainly for commercial use. What are your plans for the To B market? How does it help this year’s sales target of 1.5 billion?

Zhou Jian: Cruzr has received hundreds of millions of orders. This order is real. This order comes from various industries, including airports, banks, brokerages, and 4S stores. Our order of 5,000 units will probably be mass-produced by the end of May and delivered in early June and July. The goal of this year’s Cruzr is 10,000 units. 10,000 units, our sales are like hundreds of millions.

In addition, in addition to our big order in April and May, you will see another action of ours. We will cut into security monitoring. You can see that many of our cameras are in high places. From face recognition, it is actually There are some defects, it needs to be captured from another angle, so we will have a relatively large action in security monitoring next.

UBTECH’s recently launched commercial robot Cruzr

Q: You said that Ubisoft needs to walk on two legs, one leg is commercialized, and the other is technology. What are the commercialization directions of Ubiquit?

Zhou Jian: The layout of commercialization, STEAM education, itself is an entry point for robots to cut into life, such as education, entertainment, and companionship. These things are easy for everyone to understand. Many robot startups in China are based on education, entertainment and companionship.

In fact, I was in a meeting last night. We formed a great strategic cooperation with the world’s largest content licensor, including IP. Next, you can see our IP products, not just buying simple dolls. At the same time, dialogue and interaction, including question and answer, and robots that can be programmed, we are all doing, and it is estimated that two more will be launched this year.

There will be 6-8 Jimu robots on the Apple Store in the future, and I don’t want robot education to become a tool for college entrance examinations

Q: UBTECH’s Jimu robot entered the Apple retail store, how did this cooperation come about? What is the current sales situation?

Zhou Jian: Our cooperation with Apple is mainly on STEAM education. STEAM education, as we all know, is to really cultivate young people’s hands-on ability, including young people’s logical thinking programming ability, including some cultivation of physical and mathematical ability. I am very optimistic about the direction of STEAM, including the country’s increasing emphasis on STEAM education.

Apple is very optimistic about this field. Apple has carried out research on STEAM education very early. Apple has actually selected candidates for STEAM education around the world. In the end, the vice president of Apple came to our team to talk about this matter. It happened last October that we The Jimu robot was launched. When they demonstrated in their Apple office, they felt that this was the product they really wanted, a real robot product that could move and program servo servos, and was more in line with such a product that was promoted to and outside the school. Therefore, Apple cooperated with us from APP to final software and hardware, and launched our first product. In the future, 6-8 robot products will be sold in hundreds of Apple retail stores around the world.


In addition, we introduced Howard, and we hope to truly form a more authoritative STEAM education standard in the world. What is STEAM, and where are the STEAM standards? How to achieve the goals and requirements of STEAM education?

UBTECH’s Jimu robot

Q: Now many start-up companies are already in the robot education market, and many of them are already running various classes or collaborating with some schools. What do you think of this venue?

Zhou Jian: Everyone says that my robot can educate children and is just needed, which is really good. But in terms of education, I haven’t done any research on education before, but our company specially established “Ubiquit Education”, and carried out a lot of research and exploration on education in the early stage.

Whether it is on-campus or off-campus education, we take off-campus education as an example. Many robot training courses are actually very few companies that have really achieved good profits or really survived very well. Robot education in chain stores, in the current environment where the foundation is not very good, it is impossible for robot education to break through, or to rise further. Many parents’ awareness of robot education has not yet reached this stage. Therefore, I think that robot education should be both on-campus and off-campus in a sense, so the first step for us to step into is on-campus education. Next, you will see that our robot education will really penetrate into every school in many provinces and cities this year, including curriculum settings.

Where is the course? How does this course conform to the trend of our education syllabus, and where do the teachers come from? Now the school teachers are unwilling to give you a robotics class, there is no benefit in taking it, and he does not understand it. This is a systematic, macroscopic and very big engineering. It is impossible to accomplish this education through a simple one or two, with very little investment, so we have invested a lot in education, but everyone has not fully felt it yet. We are cultivating teachers, we are A joint venture was established with Beijing Normal University to train teachers. We are planning from start to finish. I think only when courses, teachers, personnel, hardware, and materials are all integrated into one system, it is possible to truly do a good job in school education in China in the world. Well done, then off-campus. But I don’t want robot education to become a tool for children’s future college entrance examination.

The stabilization and commercialization of servo steering gear is the key to mass production, and AI technology is fully deployed

Q: From starting a business as a steering gear, to a complete robot, to adding an intelligent interactive system, what will be the more difficult breakthrough points in the whole process?

Zhou Jian: In stages. Ubisoft is a start-up company. Our initial difficulty was the problem of the technical team. It was technically difficult for us to stabilize the servo servo, which is what we call commercialization. Maybe you can see that there are some small humanoid robots similar to servo servos on the market, but in fact, you need to stabilize and commercialize it, which is a very big breakthrough. Ubisoft has always put a lot of effort into this aspect, no matter from supply chain, technology, R&D, new product introduction, we have spent a lot of effort, which is what keeps me awake every day. Places, in a sense, we spent a lot of power.

Second, when we were in the frontier layout, everyone knew that there were not so many talents in the field of artificial intelligence, and the competition was very fierce. This is a very big challenge for startups and UBTECH, so we have been I am doing my best to not only make current products, but also make some layouts for the future.

Last year, we introduced Dr. Tao Dacheng, who has changed us a lot, and made us re-understand machine vision. How to make our robot truly understand the world, for example, when a robot enters this room, the first point is to recognize the entire surrounding environment. It is very simple for a human to enter a strange environment. The first point is to see the sofa, table and chairs in this environment. How to place it; the second step is to see how many people there are, to perform face recognition, and to know whether these people are parents, grandparents, and so on. Let me give a detailed example. When we were particularly surprised at the time, if the cooking fumes were very heavy, it would affect the robot’s judgment and changes in light, including the fumes. We thought how big the fumes could be. , In fact, it will cause the robot’s visual judgment to be blurred, and it is not as easy to adjust as humans. Then we need to atomize, remove white, and remove oil fume. These are all problems for us, so we made a lot of efforts in this area at that time. In addition, including recognizing the posture of the human body, there are many human actions, and the robot will do many actions. What is the purpose and meaning of this action? It involves multimodal visual recognition.

Because I started with hardware, I don’t have to admit that our company was relatively weak in software before, but as a start-up company, we not only want to make the current products well, so we have made great efforts in this area. layout. In the future, we hope to establish a set of global robot standards in a sense, such as what is the standard of a humanoid robot, what are its specifications, its crankshaft motion, including some basic needs of artificial intelligence, we are trying to build, but This market has not really come up yet, we are just doing some more cutting-edge and guiding work.

Q: At present, the exploration in the market is relatively good, and the business scenarios are relatively accurate. But how to plan the current reserves accumulated in artificial intelligence technology, such as image, voice, or OS? When I experienced Alpha2 just now, I found that voice interaction is still a little problem, that is, the understanding of context semantics.

Zhou Jian: As a start-up company, one is a steering gear robot company, we are facing some problems, we have speech recognition, including machine vision, etc., we have adopted this method, and this is what we have been financing. reason. In retrospect, a lot of content and a lot of technology can’t be done by UBTECH, we are now focusing on three things:

The first is the drive technology, the key technology of large-scale servo servos; the second is that we hope to build an algorithm similar to that of a better platform than Boston Dynamics, and we are very confident to do this. Because we have conducted a general analysis and in-depth understanding of the technology of all the large humanoid robot companies in the world, we will find that companies such as Google, Boston, and Honda have no way to commercialize them.

The second thing I want to consider is energy consumption. If I realize a certain scene for an hour of movement, how much energy do I need, so we put a lot of power in the movement algorithm.

The third point is that we are in the field of AI, and we talk about the brain. We prefer to focus on machine vision, because we think that most of the information sources still come from machine vision, and machine vision also requires a larger amount of data. This is also what we are laying out.

The rest of the full text page

As for speech recognition, we all know that one of our shareholders is iFLYTEK, and we will try to use it as much as we can. But in terms of semantic understanding, in fact, I think there are problems all over the world, and pure NLU cannot solve all robot problems. The future recognition understanding is a multimodal, it is based on vision, based on speech, semantic multimodal understanding. There is a company that is also trying to make efforts in this regard, and UBTECH may invest in them, take a strategic stake in them, or even cooperate more deeply with them. This is what we are going to do now.

Robot is the best carrier of AI, but it will take at least 5-10 years for the humanoid version of the robot to enter the family

Q: How long do you think it will be before humanoid robots become as popular as mobile phones?

Zhou Jian: This is a relatively big topic. There are human-like robots in robots. In the short term, it will take about ten years; in the long run, it will take three to fifty years to achieve a real human-like robot. commercialize. But in the short term, in five to ten years, some of our humanoid robots may enter the home as soon as possible. I dare not say that the mobile phone is so popular, but it is the era of active interaction center. It feels like it will come soon.

As we all know, artificial intelligence is still in a very early stage. I personally think that the most important thing about artificial intelligence is that it needs to be truly implemented, whether it is in the financial field, in the field of science and technology, or in all areas of our lives. I think robots are the best carrier of robots in a sense. This is our advantage.

Many current robot products are just transitional products in a sense. Wheeled robots, in a sense I also think are a transitional product. Only bipedal robots are designed for our human environment. However, the current problem is that the robot has not achieved real breakthroughs in driving, including joint driving, motion technology, and control motion algorithms. In fact, the biggest problem is that the driving technology has not achieved breakthroughs.

To make mass-produced and stable humanoid robots, only UBTECH is the only one

Q: Now the giants are gradually entering the field of robotics. You have just emphasized that we are a start-up company, and the things we can do are actually relatively limited. Do you think that in the future, Ubisoft will be able to use its own products. Break-even, or the possibility of living under a giant in the future?

Zhou Jian: Our products are actually divided into three parts. We have different stages from the primary basic research, to the research and development of an intermediate stage that is quickly introduced to the product stage, and then to the research and development of the product end. The thing you just mentioned is mainly in the research and development of the final product, which involves too many basic technologies. I have always described our public technology platform as a big tree. In this big tree, there are visual modules, voice modules, and servo modules used by others and accumulated by ourselves. Throw this technology into it, and then each product is on it. Raw branches, what is he going to throw this in, like a torso thing. The steering gear technology I have accumulated in the past, according to some of the accumulated technology, develop a ready-made product according to the requirements of the IP partner, this is a relatively simple thing for me, rather than touching the future. One direction, but this can already bring enough revenue for startups.

UBTECH Shenzhen headquarters is located

Q: When UBTECH cooperates with big companies on robot products, will there be a lot of competition?

Zhou Jian: We are a company with a relatively international perspective. As soon as we come up, we have a lot of potential future cooperation partners in our hands, including the formation of future technologies. We also have it in our hands, and we are ready. Let me give you an example, how many companies in China do you think can cooperate with Amazon on a Lynx humanoid robot? In fact, there are not many companies in the world that have it. When you make an IP product, how many companies do you think China can make robots? Have? Everyone is making a flat plate and wheels in a swarm. It does not have servo servos. Even if he can make a dozen servo servos, he can’t mass-produce it, and he can’t maintain stability. When the opportunity comes, only Ubisoft can get it, and we will actively promote this business. So everyone sees that what Ubisoft is doing is quite high, Apple, Amazon, and a few more famous names. I can’t say it now, this is the result of our efforts to reserve the current stage. .

For example, if you make a BB-8 for $1499, do you think anyone will buy it? Our BB-8, if I remember correctly, is $149. If he can achieve 1499 or 2499 dollars, no one will buy it. For the same reason, our cost and our advantages in the early stage are brought into play, so I believe that no matter which IP product, as long as he wants to make a humanoid thing, he will find it in the global search. UBTECH, from the perspective of mass production, scale, and stability, no company in the world can currently make the same products as us. Don’t look at this small Alpha, it is really stable and the quality is reliable. The servo servo has been used for three months, and it is not broken, only the best choice.

Ubisoft, what we are good at is to develop the real core components we have made before into various commercializations. As you can see, we have been doing it all over the world. At least so far, both Alpha1 and Alpha2, including Jimu, including Cruzr, have used a lot of servos. We have developed a lot in the core product of servos. Many product lines have brought into play our ultimate advantages.

China’s artificial intelligence and robot entrepreneurship are false, and in all likelihood they will die

Q: With the upsurge of artificial intelligence, how do you evaluate the current environment for robot entrepreneurship?

Zhou Jian: I have always believed that China’s artificial intelligence, including robot entrepreneurship, is a myth. I have always believed that 80% and 90% will die. From my personal point of view, I see many robotics startups in China have no idea what they are doing. We took something that we don’t know belongs to a tablet or something, and then put iFLYTEK or something like speech recognition semantic understanding on it. This kind of direction is good, but the fundamental direction is wrong.

Starting a business, there are several very important things: Is the market big enough, is it a pseudo demand? If the market is big enough, not a pseudo demand, OK, 100 people, 1000 people come in and do it all right, but this thing will be a transition sooner or later When the product is produced, when I do this thing just for this outlet, I think this thing is completely wrong. Therefore, you can see that a large number of such companies were established in 2013, 2014, and 2015. I think this entrepreneurial environment is actually a waste of resources, or I think this leads to a situation where many entrepreneurs will go into exhaustion. , I personally think this is not very understandable.

For my investors, for all of us, and for our colleagues, in fact, I have good intentions, and I don’t think it’s a good thing for so many startups to rush to do the same thing. You have to be your own core. For example, whether we can succeed in Ubisoft, I at least know that what I am doing is a humanoid robot in the future. This is very clear. And why be a humanoid robot? Why can I be a humanoid robot? Because I support myself, I can use more funds for financing. Why would others raise money for me? My company with such a high valuation, more than $1 billion, is because everyone sees my ability to commercialize. You have to walk a path that suits you to start a business.

The Links:   3HAC037682-001 ELECTRONIC