(ChinaIT.com News) September 10-12, 2021, the 2021 World Robotics Conference will be held in Beijing. The World Robot Conference has been successfully held five times since its establishment in 2015. It is an important platform for promoting global industrial and technological exchanges and cooperation in the robotics field. This year’s conference takes “the global robotics industry community with a shared future” as its value advocacy and showcases advanced technologies and scientific research results in various fields around the world. It aims to promote academic exchanges at home and abroad, tap the potential for cooperation, and promote the development of robotics, products and markets.
Gao Xiaorong, a professor at Tsinghua University, focuses on brain-computer interface technology, theories and methods of neural signal detection and processing, and based on this technology, studies brain function analysis methods related to biological nervous system information. The signals involved include the processing of cortical potentials and scalp EEG signals recorded at the macro level, and the processing of the neuron discharge signals recorded at the micro level and the local field potentials formed by them.
ChinaIT.com is a media platform that focuses on reporting the frontier IT technologies at home and abroad. It is committed to providing users with diversified communication and marketing services through high-quality professional content, as well as in-depth and extensive communication. At the current World Robotics Conference, ChinaIT.com, as a partner media, participated deeply in the content of the conference. Fortunately, it interviewed Gao Xiaorong, a professor at Tsinghua University. The following is a transcript of the interview.
Gao Xiaorong, Professor of Tsinghua University
Reporter: First of all, please talk about what the brain-computer interface is, its development process, and some main applications?
Gao Xiaorong:Brain-computer interface is a hot term now. It is such a technology that directly extracts signals from the brain to control peripherals. It is also a kind of human-computer interface. Traditional man-machine interfaces such as mice and keyboards require hand muscles. Nowadays, voice input uses the facial features, but the brain-computer interface skips all of these. It does not require peripheral nerves to participate, and directly controls peripherals by brain signals. Such a technology. Twenty years ago, it was still a science fiction, that is, it was a technology that “whatever you want to do”, and you just think about it and do it. This technology was first proposed by a professor in the 1970s, and named Brain Computer Interface for the first time. At that time, when the computer suddenly changed from a big house or one floor to the size of a refrigerator, they wondered what this computer could do? Can it be directly controlled by EEG? Then I remembered these and defined the word. However, when the technology was invented at that time, the technology was still very backward, so after I came up with the term, I did research for 20 years, but did not produce any results, which is equivalent to being invented again by other scientists around 2000. In fact, it was around 2000 that the first International Conference on Brain-Computer Interfaces was held. At that time, countries around the world put forward a lot of words, including 17 or 8 words such as “Thinking about it,” and finally everyone recognized BCI. This word, because this is the earliest one, so all those who use the brain to interact with peripherals are called BCI, which is such a historical process.
In fact, when BCI first appeared, the first driving force was medical driving force. We have to solve the problem of people with movement disorders like Hawking controlling peripherals to use computers. Tsinghua University also started to do brain-computer interfaces in 1997, and it was also because of personal computers. It has become popular, and we thought how to use a computer if this person can’t use a keyboard and mouse? So we thought of brain-computer interface technology, so we started this research, so the first driving force is called medical driving force. After more than ten years of development, AI was particularly hot five years ago, so the second driving force appeared, that is, the driving force of AI. The core advocates are Musk and Facebook Zuckerberg. They believe that the large-scale development of AI will eventually lead to a multi-agent society.
The original agent was just humans. Now AI can be smarter than humans. How do we communicate with computers? If we are too slow to communicate with the computer using keyboard and mouse, even if we use words to communicate, it is too slow, do we directly extract signals from the brain to “get everything we want”? So there is a second driving force. Now it is two-wheel drive, one is medical drive, the other is AI drive, one is to fill up those who can’t work, and the other is to fill normal people into superhumans who can communicate with computers. The most famous thing Musk has done is a pig. Recently, a monkey has appeared, which can use the brain-computer interface to let the monkey play games. In fact, there are several representative ones in China. Zhejiang University directly implants the brain-computer interface into the human body, and then gives a disabled person like Hawking to control the robotic arm, and can also play mahjong. Tianjin University in China puts the brain-computer interface on the space station. , Astronauts can also do brain-computer interfaces.
But from the perspective of the entire research, the brain-computer interface is still a basic thing. From a technical point of view, the brain-computer interface chip is involved, which is to extract the neural signal from the brain to make the chip, and the other is the whole set of analysis software. More than a decade ago, the research was done on software. Tsinghua was doing software. However, after the United States imposed export controls on brain-computer interfaces in the past one or two years, many Chinese companies began to do chips. Two brain-computer interface chips were exhibited at the conference, and they are both at the same level as foreign ones. The performance is exactly the same and can replace foreign ones. So this is the development of this year’s chips. In addition to chips, there are systems. Just as computers have computer systems in addition to CPUs. There are probably several domestic companies, including B.C. and Jiangsu Jicui. One is in the medical field and the other is in the consumer field. Many EEG systems have been developed. Once you have the system, you need APP applications, which are roughly divided into three parts:
- The first piece is medical treatment, how to solve the patient’s problems, and how to use it for people who can’t use the mouse and keyboard or talk.
- The second part is education. When children are learning and receiving education, they can use brain-computer interfaces to make up for their lack of attention and cognitive functions. When children should be listening in class, there will be a feedback if they are inattentive, or from these From the feedback of family members, it is known whether the child did not learn because he did not listen to the lecture, or did not learn because he did not understand. These are two big problems. If you don’t understand what you haven’t learned, it’s a logical problem. If you don’t listen to the lecture and you didn’t learn it, it’s another problem.
- The third piece is related to industry, that is, industry 0 and so on. In fact, the core issue involved is safety. The easiest example is autopilot. Your car can be autopilot or manned. How do you coordinate when people drive and when autopilot? Two people can’t fight there. I want someone to drive. He wants to drive automatically. How should I switch? If there is a brain-computer interface, it can be easily done, such as inattention, drowsiness, drinking, etc. Vehicle detection is not suitable for human driving, then the vehicle will drive automatically. If it can’t drive, it will automatically stop on the side of the road. It can prevent accidents. For example, when making a phone call, the vehicle can stop when it finds that it cannot continue to drive. Sometimes people get sick and suddenly have a bad heart. If the vehicle is found, switch this thing. People can switch when they find that the path chosen by the car is wrong. Therefore, there is a safety issue in Industry 4.0.
Reporter: What is the current direction of brain-computer interface research? What are the technical problems that need to be solved?
Gao Xiaorong:There are several approaches to brain-computer interface research: one is invasive, which is to put electrodes in the brain, and the other is a non-invasive method, which is to put electrodes on the scalp. Invasive methods are more informative, but people may not be able to receive them. It is difficult to insert an electrode into the brain properly, so use a non-invasive method to implant electrodes on the scalp. It’s like putting the microphone inside and outside the house. The signal is much weaker outside the house. There are many methods of information detection, so this is weaker. If you can hear loud voices inside the house, or someone laughs outside the house, But being able to know what is laughing in the house is such a process. Musk’s research is all invasive methods. It is actually difficult to put motors into the brain nerves, because brain tissue is the softest tissue in humans, except for body fluids, which is the softest brain tissue. The silicon we monitor is the hardest thing. Putting the softest thing and the hardest thing together, you can imagine that there will be problems in the working environment. It’s okay if two people don’t move at all. You have to do some exercises, and if you have a jump, the blood of your heart will fluctuate, so how much is the loss of soft and hard? The difficulty is still great, the first difficulty is the physical difficulty.
The second difficulty is the biological difficulty. Even if the stuff is soft, it is toxic and harmful to the human body.
The third aspect is information. The human brain has 100 billion neurons, and each neuron is working. According to the bit rate of the communication rate, this is at a speed of 1TB per second. When we communicate today, if we sample my language, one hundred bits can transmit my language to you, and when you hear one hundred bits in your ears, it will be re-interpreted into 1TB, so this is based on the language-based compression capability. This information problem is solved in the process of human evolution. A hundred bits is still not enough if compared with a computer, so this is a very difficult thing.
Why do we need to study such a difficult thing? Why not just not do it and wait for AI to do it? But in fact these are impossible, because AI can only solve the problem of fact, that is, what is this thing, but it does not solve the problem of value. I mean social value in a broad sense. For example, if this is a bottle of water, AI can recognize it immediately, but the value of this bottle of water is not known to AI, because the value is related to the environment. Now water is a value here and another value in the desert. If Human beings are short of water, and when the entire water resources are depleted, it is another value. All these value issues can only be judged by humans. Machines can judge them, and we cannot impose them. The same thing is viewed by different people in different values. For example, in the story of Lu Xun’s human blood steamed buns, there is definitely a problem with killing people, but that one thinks that it is valuable, so people and people look at the value completely different, so we must have this value.
We assume that AI develops infinitely and grows bigger and bigger. How can we maintain human dignity? A dialog box pops up to let you choose Yes and No. If the AI is playing anytime, clicking Yes and No all day will make mistakes, and it will be annoying. Why do you ask me for such a simple thing? But if AI does it for you, you may not do it. Just like a child, we must maintain the dignity of human beings. We must let the machine understand our value. It depends on whether we want to do these things. Brain-computer interface is the most important thing. It is a human-computer interaction tool. We will never tolerate the emergence of AI to replace humans. If we find that the future development of AI will definitely eliminate humans, we need to eliminate AI now. Humans and AIs require coexistence and development, and we have to take the lead.
Gao Xiaorong, Professor of Tsinghua University
Reporter: Is there a plan for the brain-computer interface in the next few years? What kind of application efficiency do you want to achieve?
Gao Xiaorong:The first thing that can be implemented now is medical treatment. Medical treatment can be clearly stated. During the process, eggs may be laid along the way. Then, medical technology can be developed to do some cognitive and decision-making things. How do people do when we interact with machines All these need to be done step by step, but now the only need is medical treatment. In addition to education and safety, other application ethics must be restricted. For example, with a brain-computer interface that improves the governance of the family, what about these people who are useless? But there is no ethical relationship in medical care, and any development will definitely be restricted by ethics. However, this technology is an indispensable technology in the future. It is necessary to develop this basic technology. In the future, there must be a brain-computer interface to realize the advantages of humans and machines. The brain-computer interface is a wide-ranging field. The materials involve physical aspects, as well as people in neurocognitive science and information science who need to communicate. When we do it, we are facing people in the medical field. We It has cooperation and research with Tiantan Hospital, Xuanwumen Hospital, Changzheng Hospital and Union Hospital.
Reporter: The brain-computer interface that Musk said is still quite advanced. It may enter the era of normal people or even superhumans in the future. You were more cautious after talking about memory access and memory sharing before.
Gao Xiaorong:Yes, I still hold this view, because the brain-computer interface is indeed something that has a very wide range of influences. If it is harmful, when you enter the human brain, you don’t know what you might get, but you know that you have lost it. What things, you must be careful at this time. For example, implanting something into the brain, a very hard object and a soft object, imagine that the movement of the brain tissue is there like a sea wave. You take a sharp object and shake it several times, so this kind of thing must be ethical first. See clearly that there is no harm to people, or I can accept the harm, but I can’t accept it in case the harm occurs, and the benefits are not obtained. After Musk comes out, he will also have to make medical applications first. It is impossible to say that in order to let people play games, put an interface in their heads. This is not possible, but those who cannot speak can use this to speak. When we have overcome all these difficulties, including physical problems, biological problems, and information problems, we will find a way to apply them.
Reporter: As China enters an era of longevity, will your research direction and research model change?
Gao Xiaorong:The brain-computer interface is very concerned about the problem of the elderly, that is, the function of the elderly is attenuated. Is it possible to use something to help them, for example, if the eyes are obstructed and the way is not visible, the brain-computer interface can be used to help remind some things, so the old age is also a problem Medical issues.
Reporter: I just heard you talk about a lot of technical aspects. Let’s talk about this conference. It is the first time you have participated in the World Robot Conference. How do you feel? Does the conference promote your field?
Gao Xiaorong:In fact, we had cooperation at the beginning of the Robotics Conference. At the beginning, we did not mention the development process of China’s brain-computer interface. The development of China’s brain-computer interface began in 1997, and by 2010 it has begun to be recognized by the scientific community. At that time, everyone published scientific papers, mainly papers. The National Natural Science Foundation of China asked us a question, and we received many, many fund applications. Because this is a very new field, we don’t know who is good and who is not. Can you evaluate it from the perspective of application without a paper? We did the first China Brain-Computer Interface Competition in 2010. It’s actually very simple. When the NSFC supported it, we applied for all NSFC topics. We invited all those who applied for it. We have typing tasks, There are car control tasks and everyone’s platform is the same. You can compare and see who is better. The first brain-computer interface was a dark horse. It was said that it was Professor Li Haiqing. He heard this story as soon as he returned to China. He came to Beijing in short sleeves. He didn’t know the temperature difference in China was such a big difference, but he performed very well at the time. Later, the NSFC saw it. He might not be able to see it during normal review, but he performed very well on it. We can also see who is the first and who is the second through the competition. A lot of information is given to the NSFC. , Which will help them guide the choice of topics.
We did the first and second brain-computer interface competitions. It was 2015 at the time. In 2017, there was the World Robotics Congress. Then they told us to do brain-computer interface because it was a broad concept. , We did a skill competition on the platform of the World Robot Conference, just to answer how many people are suitable for brain-computer interface, and whether they can use brain-computer interface. Because at that time, all the people doing brain-computer interfaces were our students and young people. How many people could use it? No one in the world knows about this. Just like when we build cars, how many people can drive? Are we going to set up a driving school for everyone to learn brain-computer interface first? We don’t know at all, so we made a skills competition, which is equivalent to a brain-computer Olympics. There is a typing and a driving car, and you can sign up as you want. It will be the second in 2018. At this time, we got the world’s largest model population, which used to be dozens of people, but now we are hundreds or even thousands of people.
In 2019, the technical race and the skill race were combined into one. The technical race produced the best car, and the skills race found the best racer. So that year, the brain-computer interface world record appeared. Take a look at the best. How fast the person can run with the brain-computer interface, that person typing with the brain-computer interface is faster than typing with the keyboard, and is also the best. It was selected from thousands of people. Because of the epidemic in 2020, we only did a technical competition. This year is the technology competition and skill competition. It is considered as the fifth China brain-computer interface. Through the World Robot Conference platform, it has greatly promoted China’s brain-computer interface, and it has also had a great impact internationally, because it is indeed Target a very large crowd. We were adults in 2017-2018, and there will be youth competitions in 2019. It depends on whether children can use brain-computer interfaces. This year, we have done 100 elderly people over 60 years old to see if they can use it. I found that the elderly can also use it. If you can’t develop a technology, it can only be used by young people. If you get rid of the elderly, and finding one or two elderly people can’t explain the problem, we did it over 70 years old, and the oldest one is 81 years old. .
This year we launched a group brain-computer interface on this platform, which is also the first one in the world. We communicate together through brain electricity, without language communication. We call it God Chat System, which supports up to 20 people, but it’s okay to support 40 people. , Is everyone chatting together. If supporting twenty people is equivalent to a classroom, everyone can exchange knowledge without speaking, or the classroom can know who has attended the class and who has not, this problem will be solved. Our brain-computer interface has been supported by the Institute of Electronics and the World Robotics Congress since 2017. In fact, these two chips are also broadcast live according to the competition system, which is the chip we launched in this competition. This game is very fair. If different people do the same problem, the result will come out immediately. The brain-computer interface is this feature, and the result is real-time. So after the competition, we did the brain-computer interface for the first time. The Fund Committee said that there were a lot of complaints in previous competitions. , We are also watching, others are watching, because it is real, all brain-computer interface manufacturers have shown their own things at this conference.
Reporter: Your research fits well with the four major state-owned industries. From your perspective, let’s briefly talk about your feelings.
Gao Xiaorong:The original development of the brain-computer interface was driven by interest. In fact, all research is driven by interest research. As Musk has improved everyone’s awareness, many companies also want to do this. Everyone in all our laboratories They can come, just because they can’t come because of the epidemic. Now is to let these companies understand what the brain-computer interface is, I think it is similar to the interface of industry and scientific research, and there are also many good companies in Yizhuang after finishing here, everyone can have a mutual exchange in this link. Platform.
Reporter: What are the major achievements of your research group in the past two years?
Gao Xiaorong:China’s position in the field of brain-computer interface, we accounted for about one-third, including the number of papers and contribution to the world are one-third, but my country’s brain-computer interface funding is one-tenth compared with the United States It’s not there, so I still use relatively little investment. The brain-computer interface for steady-state recognition and induced positioning is to put frequency markers into the brain-computer interface. The brain-computer interface proposes a total of more than a dozen paradigms, and now there are only three recognized paradigms: SIP is one kind, and imaginary movement is One, K300 positioning is one, and the fastest brain-computer interface in these three paradigms is SIP. This year we published a very important article that summarizes the development path of brain-computer interfaces and proposes the I3 model. The I of BCI is actually three I: the first I is the Interface interface, which is one-way. The second I is Interaction, which is a two-way thing. The third I is Intelligence. The brain-computer interface will finally evolve according to these three models, and finally it can reach the level of brain-computer intelligence, that is, after the combination of brain intelligence and machine intelligence, humans can do it. This is proposed by Professor Gao Shangkai, the former director of brain-computer interface research at Tsinghua University. He is my teacher and proposed such an I3 model.
There is another contribution recently. If you want to see it later, there will be a youth essay contest this year to show the result. It turned out that we had 6 goals for brain-computer interface control at first, then 12 goals, and then 40 goals. This year, 160 goals have appeared. We are now a group multi-person brain-computer interface, and we have also made contributions in this regard.
Reporter: What is the current application situation of the technology you mentioned just now?
Gao Xiaorong:We have such applications as controlling manipulators in Tiantan Hospital. “Challenge Impossible” has a Chinese Gradual Onset syndrome. When we were doing clinical practice, we found that such a patient could type in with a brain-computer interface and participate in Dong Qing’s “Challenge Impossible” program. , And then recite a poem with Dong Qing. This person is already speechless. The ventilator is plugged in, but his brain works well.
Reporter: So from a global perspective, what are the breakthroughs in brain science and cognitive disciplines?
Gao Xiaorong:The brain-computer interface and cognitive science are actually developed in parallel. The brain-computer interface is equivalent to opening a window of the brain, which can help the development of neuroscience. The new research results of neuroscience in turn promote the application of brain-computer interfaces. The two are not in tandem, but develop in parallel. It’s not that the brain has developed something. We then develop the brain-computer interface, so this is parallel. With an emotional brain-computer interface, we don’t know what is going on with emotions, and we can adjust when we detect whether the emotions are good or bad.
Reporter: We talked about the challenges of brain-computer interfaces. For researchers and technicians, how to deal with some real problems?
Gao Xiaorong:We must balance our goals, actual capabilities, and methods, and not let it go too fast. When you see it, everyone will study it in a swarm and remove all ethical restrictions. It will be troublesome if something happens. This must be comprehensive, involving the coordinated development of various disciplines, and progress without other disciplines is also limited.
Articles on the ChinaIT.com website are limited to providing more information and do not represent the standpoint of this website. For reprint, please indicate the source. The reprinted article comes from the Internet. If you have any copyright issues, please contact us: content@chinait.com.
.