Robots In The 5G Wave: Technological Iteration And Industrial Evolution To Speed Up The Pace Of Autonomy

In September this year, a video from the other side of the ocean detonated the technology circle. Boston Dynamics's humanoid robot has been able to smoothly and steadily achieve forward somersault + 360 degree jump.

Although this looks like the existence of "dazzling technology", the development of robot industry behind it is worth exploring.

In the recent "GBAS 2019 Bay robot and AI conference", Yang Huayong, a member of the Chinese Academy of engineering, told the twenty-first Century economic news reporter that the core technologies of Boston Dynamics include two aspects of 3D printing and key components. From the Chinese point of view, the independent development of the key components is still on the stage of climbing. Although some big power devices have reached the forefront of the world and have achieved the world's leading position in individual fields, many key components still face short board.

"The next 15-20 years will be a critical period for China to become the world's manufacturing powerhouse and make up the short board on key components." Yang Huayong continued.

Of course, there is no lack of boosted utility of basic core technologies such as AI and 5G. Xi Ning, Professor of robotics and automation at the Department of industrial and manufacturing systems engineering at University of Hong Kong, told the twenty-first Century business reporter that the application of intelligent robots in manufacturing industry faces 3 new challenges: programming, correcting methods, and integrating robots with sensors. Solving these problems is an important step for industrial robots to meet the needs of intelligent manufacturing.

Robot problem is fundamentally a communication problem. Mobile communications, especially the new opportunities brought by the fifth generation of mobile communications, provide excellent facilities for AI applications. Xu Xiaohui, vice president of Dawei technology, pointed out to reporters that the arrival of 5G is undoubtedly a revolution for the rapid development of intelligent robots.

Intelligent robot short board

Ding Han, academician of the Chinese Academy of Sciences, proposed that robots will go to "communion" many years ago, which means that robots and environment, robots and human beings can all adapt highly to the three. This is the only way for future robot development.

However, he told the economic news reporters in twenty-first Century that in this direction, the progress of the industry is quite different now. This is due to the fact that cloud computing and AI are the wings of intelligent robots. But to the development of "communion", new materials, sensing and driving technologies are needed. "The core of communion is the rigid structure of robot body, which is driven by a generation of materials, technology and equipment."

Ding Han emphasized that the development of Boston Dynamics has gone through a long time of technological accumulation and has reached the commanding height of technology in the world today. Its products have reached a near perfect level in terms of dynamics control, which will play a leading role in the development and promotion of robot technology in the future. But whether it can be converted into products that can popularize the industry or household, we need to further explore the landing mode.

Yang Huayong, academician of the Chinese Academy of engineering, told the twenty-first Century economic analysis reporter that the core technologies of Boston Dynamics include two aspects of 3D printing and key components, which has led to the miniaturization of components such as hydraulic digital valves.

"Boston Dynamics has the world's top stability team to promote development, which is a reference for China's R & D system and basic application research." He further pointed out that, in particular, local governments should not go too far in pursuit of hot spots, instead of investing in more forward-looking areas and providing sustained and stable support for technological development, thus forming brand effects and gradually finding and occupying vertical markets.

In the early years, Yang Huayong took part in the process of autonomous development of shield machines in China. He pointed out to reporters that in fact, the basic components often walk in front of large equipment. If there is no accumulation of components, it is very difficult to realize the autonomy of large equipment. This requires the promotion of basic technology and industrial chain, and gradually solves the problem of consistency and reliability, and then considers mass production and price.

"The more we go upstream, the more industry and technology accumulate, the greater the amount of investment. In terms of key components, we need to occupy the market on a large scale, and we need brand, word of mouth, and long-term stable quality support. This is not achieved in the short term. Yang Huayong added that the development of key components is related to basic technology, basic materials and basic software. The next 15-20 years will be a critical period for China to become a world power in the manufacturing field, because most of the short boards will be gradually supplementation during this period.

For the current process of intelligent transformation of China's industry, Yang Huayong pointed out that at present, some process industries have realized the importance of intelligent transformation, and accumulated more data in this field, so that they have achieved good results. The discrete industry has a huge number in China, but it also develops unevenly. Whether the enterprise hesitates to carry out intelligent transformation is the core of how to support its cost, which requires the product to get rid of the low label and take the high-end path. "Enterprises should take part in the subdivision from bottom to top, first break the subdivision links, and gradually realize the overall intelligence, and then consider creating platforms. Cloud computing is only the first step. When it comes to global data governance in the future, the phased effect can be calculated.

AI technology iteration

In the process of landing robot development, AI is undoubtedly a good companion with its common growth. But in the era of weak artificial intelligence, how to break through the technical barriers and get better landing is a technical problem currently facing.

Even if Google is widely used in AI technology, its intelligent voice assistant is still indispensable for manual assistance.

To this end, deep awakening technology co founder Yuan Peijiang pointed out to the twenty-first Century economic report reporter that deep learning in the 2 period is facing the bottleneck of data saturation and ability approaching to the limit, and its shortcomings will become more and more obvious. This means that relying solely on the training pattern of annotation data in the past will not be enough. Let alone annotated data, there will still be noise and error.

"When the model depends heavily on training data, if the data itself is wrong, the conclusion is likely to be wrong. This is the third generation of AI that we have to face. He pointed out.

So artificial intelligence in the future may not be based on the current black box and deep learning structure, but a kind of explanation, believable and robust. On the basis of existing ones, we need to face the collaborative innovation of artificial intelligence and big data.

Yuan Peijiang said, this is also the core direction of the company in exploring the 3 version of the deep learning model. It is also important to consider that the security mechanism of AI needs to be optimized continuously. "When we reach a certain stage, turning back will strengthen many technologies. This is the only way for the development of technology and industry.

The current AI capability is not enough, but in the next 5-10 years, this technology is bound to be an indispensable partner of mankind and will bring great changes to human life.

Yuan Peijiang continues to say that the most widely used field of AI technology is vision and speech, because of the large amount of data accumulated in these fields. There are many demands in the industrial scene, but finding pain is not easy.

"Personally, it is believed that industrial manufacturing will bring new opportunities in the process of gradual development, as well as trial and error. From the current stage, the volume of the current stage of development is not enough to support the growth of enterprises in this field as a giant, and a relatively challenging market. Of course, any technology will eventually find the most suitable soil growth. " He concluded.

In addition, the traditional industrial robot programming is time-consuming and complex, but with the current 3C industry product update cycle shortened to 6 months, the complexity of programming has brought a high demand. "Our train of thought is based on product design, so that the computer can automatically analyze and design the robot through AI, and automatically generate robot programs." It can reduce manual programming and test programs faster.

Because the robot itself is equipped with sensors, the real-time information acquired through sensors is combined with natural language technology, so that the program can be understood faster and the robot can be programmed quickly.

5G leads industrial transformation

The evolution of the underlying communication technology will further bring about great changes in the robot market.

Ning told reporters that 5G has the characteristics of low latency, large bandwidth and high reliability, which provides a new means of communication for robot applications. "So the future will be a more complex and large-scale coordination system, which can significantly improve efficiency and accomplish more complex and intelligent requirements."

For example, under the 5G technology support, robots can help humans expand the functions of hands, eyes and ears, assist remote machining and manufacturing, and overcome the problems of intelligent manufacturing development caused by distance, scale and environmental constraints.

Yuan Peijiang thought, "the rise and fall of AI technology in early years is that technical conditions do not have enough support. But 5G and the current second generation deep learning model will bring enough scenes and resources, which means that 5G and AI will accelerate each other's development.

Tatsu technology has launched a cloud robot product. Xu Xiaohui, vice president of the company, analyzed the twenty-first Century economic report reporter. There are 100 billion neurons in the human brain, but to put on the robot to achieve a strong computing power, it may take a ton of unit weight of the robot brain to achieve.

This means that cloud is the inevitable result of large-scale collaborative operation of robots, and 5G is the catalyst. "We can say that cloud intelligent robot is the killer application of 5G." He continued that the robot's individual learning is difficult to get a systematic ability to upgrade. Through cloud connectivity, the robot becomes a group learning, which will further enhance the robot's ability.

The hidden threat that this may bring is the escort of block chain technology. Xu Xiaohui further explained that the functions of robots in the future will be much more than that of mobile phones, so it is necessary to guard against potential incidents of machine injury. Using the mutual trust mechanism of block chaining and attestation and control of robot operation instructions from cloud, we can solve this problem.

"This will bring about incredible changes." Xu Xiaohui concluded that the cloud side intelligent robot is developing to the robot ontology -5G Network - edge computing node - robot safety private network and large-scale cloud brain. When the amount of access is increasing, the data processing and transmission of cloud will bring fundamental changes. Robot cloud will be the only way for the industrialization of intelligent service robots.