Shulou(Shulou.com)06/03 Report--

From July 12 to July 14, the fourth Global artificial Intelligence and Robotics Summit (CCF-GAIR 2019) was officially held in Shenzhen 2019. The summit is a top exchange event in the three major fields of artificial intelligence and robotics academia, investment and industry, and wants to create a powerful platform for cross-border communication and cooperation in the field of artificial intelligence. The summit was hosted by the China computer Society (CCF), hosted by Lei Feng.com and the Chinese University of Hong Kong (Shenzhen), and co-hosted by Shenzhen Institute of artificial Intelligence and Robotics, with strong guidance from the Shenzhen Municipal Government.

After the meeting, Lei Feng was honored to have an exclusive interview with Professor Chen Yiming, academician of Singapore Institute of Engineering, professor of Nanyang University of Technology and chair of ICRA 2017. Professor Chen gave his unique views on some recent developments and development of artificial intelligence and robots, and put forward a lot of valuable and practical suggestions according to the current research situation of contemporary academic youth.

The following is all the wonderful contents of this exclusive interview. Lei Feng's AI Science and Technology Review has been sorted out and edited without changing its original intention.

Chen Yiming

IEEE Fellow&ASME Fellow

Fellow, Singapore Academy of Engineering

Professor, Nanyang University of Technology, Singapore

Former chair of IFToMM Robotics and Mechatronics Technical Committee

Professor Chen Yiming's main research areas are robotics and automation, including robot design methodology, precision actuators, wearable sensors and interaction. In addition, he is involved in logistics robots and infrastructure robots. At present, he has published 115 journal papers and 224 conference reports, published a total of 5 works, and owns 5 patents.

Where on earth do construction robots and logistics robots go?

Lei Feng net AI Science and Technology Review: what is the recent development of construction robots and logistics robots?

Chen Yiming: from the perspective of society as a whole, some companies have begun to implement the technology of construction robots; at the same time, the Singapore government is also strongly supporting the research and development of construction robots.

Here need to emphasize a concept, building artificial intelligence is far from as simple as imagined, it is not only robot technology, it also includes automation technology. Because building is a very complex project, the whole project needs the cooperation of different links, such as process improvement, equipment optimization and construction method innovation, and so on. Only when each link is optimized and improved, can it have a positive impact on the whole construction process.

At present, the construction robot belongs to a systematic project. Some of these parts have developed earlier and are relatively easy to do, and the technology of this part is mature and complete, while some of the more difficult parts of the government continue to support us to do in-depth research. Therefore, the whole construction robot project is still in a state of slow development, and whether or not the concept of "smart city" is realized in the future, the system engineering of construction robot still needs to be optimized and improved.

In terms of logistics robots, the earliest companies to use the first generation of logistics robots are Amazon, ATV manufacturers, Ji Zhijia and so on. However, at present, logistics robots are already developing to the second generation, which will mean that the whole logistics robot will enter the next track-the era of strong combination of software and hardware.

This is because at the beginning of the development of logistics robot, people's understanding of its core is the creation and application of hardware; now, the demand of the whole society for system software begins to increase. We often do not pay enough attention to intellectual property rights, so many people think that software is easy to obtain, so we ignore the importance of software; but this is not the case, often some large-scale system software is very difficult to obtain, such as some industrial-level software, including high-level software such as CAD are not easy to obtain. And it should be noted that the robot itself is a complex system. Hardware is easy to be plagiarized and copied, but software for different needs is difficult to replicate. Today is the era of the combination of hardware and software, so the research and development battlefield of the second generation logistics robot has just begun.

Lei Feng AI Science and Technology Review: according to relevant reports, Professor Chen's team has landed an example of logistics robot in April this year. Can you talk to us about the challenges and opportunities encountered in this process?

Chen Yiming: in April this year, we applied the technology of logistics robot in intelligent warehousing in Yiwu. Through the combination of our technology and the appropriate landing scene, the intelligent robot system solves the problem of small and medium-sized e-commerce warehouse and logistics employment.

The reason why this technology can be transformed into commercial value and achieve landing is that we have found a commercial landing scene that matches the technology-Yiwu. Yiwu is the national Taobao e-commerce capital, but also brings together the major three cloud warehouses, so its entire transportation cost is low, it can be called the paradise of all e-commerce. In this large background environment, it is very conducive to the landing of our robot technology. In the process of technology research and development in the future, the choice of landing scene is also worth thinking about.

However, although the scene problem has been well solved, the actual cost of the landing process can not be ignored. How to enable small and medium-sized e-commerce companies to overcome the problem of technical cost is a great challenge. However, fortunately, the local banks in Yiwu have a certain degree of support for the business financial model, which is another key to our final landing. It is worth noting that the mode of business operation will largely determine the effect of technology landing. In this process, the bank's support to the merchants will be conducive to the combination of the whole technology and products. And this kind of support from policy can also be seen as a great opportunity for robots in the process of landing.

Lei Feng net AI Technology Review: what is the change and impact of the strengthening of digital model construction in the Chinese market in the past two years on the transformation between construction workers and robots?

Chen Yiming: in the process of building a digital model, the big problem is that there are many rules and cumbersome procedures, so the cost is very high. This is unaffordable for most companies, including the Singapore government, which has been implementing it, with little effect.

Therefore, in order to really achieve digital model construction, on the one hand, I think we can add policies similar to national demonstration projects in some of them, from simple construction to complex construction, forming a gradual process. On the other hand, we can also adopt appropriate labour policies to control the quota of foreign workers through the Government, so as to strengthen the follow-up of machine automation. This is precisely the method currently adopted by the Singapore Government.

Of course, when implementing policies, different policies are needed for different industries, which requires us to make an accurate judgment of the industry when adopting policies. Therefore, the core of the replacement of human work by machines lies in the balance between the unit price of manpower and the unit price of machines in the industry, which will determine the final implementation effect of the policy. Judging from the 30-year iteration of robots, robots are currently in a period of third-generation change, and I expect this period to last at least ten years, so robots will not have a great impact on the entire economic market for the time being.

Professor Chen Yiming is at the "round table of the Chinese top meeting".

Lei Feng AI Technology Review: at present, there are also some innovative companies that do logistics AI. Compared with some large enterprises, what are their advantages or breakthroughs?

Chen Yiming: as far as I know, although the logistics AI market is not yet fully mature, it has slowly entered a cold window. This is because the current era is more in need of mature and complete solutions, and a single technology has been unable to meet the needs of the market.

Therefore, I think the whole market threshold of logistics AI may be higher and higher. Small companies want to seek breakthroughs unless they work together to expand the market. However, in China, many small and medium-sized enterprises are squeezed by large companies, which is not only a great harm to small enterprises, but also not conducive to the development of the whole industry. Germany is just the opposite of ours. Many of their small and medium-sized enterprises blossom and operate at low costs, which has created Germany's status as an industrial power in the world.

So should the development of robots. The efficiency of large enterprises is often not the highest, but focus on a certain vertical field of manufacturers, their core competitiveness may be stronger. This view has also been well verified in the export transactions of Germany and Japan. Including the United States, in the overall situation, there are not only large companies, but also vertical small companies, they form a good healthy competition with each other. In my opinion, the real maturity of a market can not only accommodate large companies, but also support the development of small companies, large and small companies work together to form a grand blossom. In our country, there are many large companies in charge of the overall situation, which to a large extent led to the current situation that the core competitiveness of enterprises is not high.

Lei Feng AI Technology Review: so for small companies, the development of fully vertical products will be a better choice?

Chen Yiming: if a company wants to be based on the world market, I think it would be a good choice to adopt a completely single vertical route, and then its competitiveness will gradually become the strongest. In fact, most of the companies that make mobile phone parts in China operate in this way, but it is difficult for us to see small companies from media reports, but on the whole, these companies are profitable.

Lei Feng net AI Science and Technology Review: in view of the polarization of large and small industrial companies in China, is there any solution, or a direction worth thinking about?

Chen Yiming: rather than regard this phenomenon as a problem, it is actually an opportunity.

For example, many small and medium-sized enterprises in Taiwan and Japan benefit from the solutions proposed by the policy that are in line with small and medium-sized enterprises. As mentioned before, vertically developed companies sometimes happen to be able to solve the pain points of cost unit prices of some large companies. Therefore, what a small company needs to do is to find the pain point. as long as he focuses on doing the simplest things, he is also very competitive.

An overview of the social ecology of robots and artificial intelligence

Lei Feng AI Science and Technology Review: in Professor Chen's mind, how do you define artificial intelligence and robots? What kind of relationship is there between the two?

Chen Yiming: first of all, we need to know that artificial intelligence itself actually contains many categories, such as pattern recognition, artificial recognition, natural speech processing and so on. In this sense, it and the robot are independent individuals. When the technologies of both are very mature, they complement each other and produce a close and powerful combination, such as the close relationship between machine vision and industrial robots and service robots. On the contrary, if one of the technologies is not mature enough, it is difficult to combine the two closely together, such as the status quo of natural voice processing and social robots, which are still in the process of development.

Lei Feng net AI Science and Technology Review: when it comes to the combination of robots and artificial intelligence, intelligent medicine is developing like a raging fire. I wonder if this belongs to the close combination proposed by Professor Chen.

Chen Yiming: in terms of artificial intelligence in medical treatment, our laboratory will develop some medical devices according to actual needs. We have done research on logistics handling in hospitals, such as automatic work of drugs, cleaning of medical devices, and so on. These are all based on the actual situation. Because in hospitals in Singapore, it is very difficult for professional nurses to clean various tools while taking care of patients. In such a scenario, the instrument handling automation tools we make can well help to solve this problem. Of course, the hospital patients' sheets and pajamas are handled similarly, how to clean, how to transport, how to organize, can be handled by the robot.

But we don't make medical robots that have direct contact with patients. On the one hand, the research and development cycle of such robots is very long and requires a lot of financial support, which can not be solved by our laboratory; on the other hand, the market demand for this kind of robots is very small, and before officially entering the market, there will be a very tedious series of certification processes, these problems are some of the major challenges that exist at present.

Lei Feng AI Science and Technology Review: so, which area do you prefer the combination of artificial intelligence and robot?

Chen Yiming: personally, I am more optimistic about manufacturing. Because from the perspective of labor cost, the application of artificial intelligence and robot in the manufacturing industry can well balance the relationship between labor cost and production value.

As we all know, in the manufacturing process of Industrial 4.0, the value of workers is mainly reflected in the quantity and quality of the products produced; and in the current situation of continuous increase in labor costs, for manufacturing enterprises, it may be a better choice to use robots instead of manpower for labor.

Lei Feng net AI Science and Technology Review: the Industrial 4. 0 you just mentioned is actually what many enterprises are doing, so what are the core difficulties that enterprises are facing in this process?

Chen Yiming: as far as I know, the core difficulty of these enterprises in the process of implementing Industrial 4.0 is that the whole project volume is too large and fails to meet the needs of customers, so most of them will stop at a certain stage.

Because for enterprises, the products can not meet the needs of customers, which will lead to the imbalance of the industrial economic chain, especially some small and medium-sized enterprises, their economic strength will be difficult to complete a long-term and costly project. For some large enterprises, the process of Industrial 4.0 will also be a protracted tug-of-war, in which the technical problems are still beset with difficulties.

Lei Feng AI Technology Review: what do you think of the current competitive environment in China's robot market? Is it a bubble or a healthy competition?

Chen Yiming: in fact, the robot bubble is a common phenomenon all over the world, but in some places it is more serious, and some can barely control it. But I think the robot market still needs to be viewed according to sub-areas, such as robot manufacturing companies, most of them are very strong, and these companies have their own core technology. And some areas that are not rigid demand, such as educational robots, these may gradually withdraw from the market.

But many things can not be judged only by the surface. A company that makes toy robots in the United States once received 200 million yuan in financing, but eventually collapsed, and some people may think that this is a robot bubble. But looking into it, I think there may be something wrong with the company's finance. The company spends a lot of money on the introduction of talent, but neglects the quality of the machine itself. Among them, there will be many factors that lead to the collapse of the company.

Lei Feng AI Technology Review: do you think the advent of the 5G era will have an impact on the design of robot technology?

Chen Yiming: I think there are two main effects of 5G on robotics. First, the improvement of 5G speed will be conducive to the testing and application of some outdoor robots, such as intelligent driving, field robots, etc.; second, the opening of 5G can support the separation of software and hardware, which will be greatly conducive to the protection of software. and change the existing business model. As mentioned before, in this era, robots need a combination of hardware and software, at this time, if some enterprises blindly copy the hardware design, but can not get the software, the function of the robot can not function properly. So to sum up, I think the advent of the 5G era will have a lot of interesting and positive effects on robot technology, and I am also looking forward to another breakthrough in robot technology in the 5G era.

To solve the puzzle of academic youth

Lei Feng net AI Science and Technology Review: there are many students in our AI study club who seem to find it difficult to distinguish between industry and academia, so what suggestions do you have for their future study and development?

Chen Yiming: see more, listen more and do more. Now many schools have internship opportunities, whether undergraduate, master, doctor, students can learn about the current industrial state of society through internship. Of course, visiting the factory is also a way to understand the needs of society, which will help students to have a better grasp of what they have learned.

However, this also needs to be considered in the light of the actual situation in different fields, such as pure physics and pure mathematics, where they can do research calmly; but engineering students need to participate more in social practice, come into contact with some challenging problems in industry, and think about solutions independently, which are conducive to the cultivation of students' innovative ability.

Secondly, we should learn to find problems on our own and do academic research spontaneously. For my students, I usually ask them to find the direction of the research project by themselves. This can not only help students to understand their own interests and future development, but also help them better adapt to social development in the future.

For another part of the students who are knowledgeable but still confused, it is very important to communicate with the mentor in time, and the mentor can have more mature advice and guidance on your research direction. Therefore, be sure to maintain good communication with your mentor.

Https://www.toutiao.com/a6719754455224451596/

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.