Integrated circuits, AIGC, robots... 7 Fudan scientists foresee the future this way

【Editor's Note】

May 27, 2023, is the 118th anniversary of Fudan University. “All kinds of activities during the school anniversary are centered on promoting scientific research.” Ever since the then president Chen Wangdao put forward this idea on the eve of the school anniversary in 1954, holding scientific lectures during the school anniversary has become one of Fudan’s important academic traditions.

Continuous academic tradition, hundreds of years of endless singing. Starting from May 9, more than 50 famous Fudan teachers from various disciplines of liberal arts, social sciences, science, engineering, and medicine will give academic lectures one after another.

In the new engineering special session "Continue to Promote Integrated Innovation and Create New Momentum for the Development of New Engineering" to celebrate the 118th anniversary of the founding of the school. Window to the future of technology.

Liu Ming, director of the National Key Laboratory of Integrated Chips and Systems, dean of the Institute of Chip and System Frontier Technology, academician of the Chinese Academy of Sciences, academic leader of the National Key Laboratory of Photovoltaic Science and Technology, deputy dean and professor of the School of Information Science and Engineering Zhan Yiqiang, Professor of the School of Computer Science and Technology Yan Bo, Deputy Director of the New Generation Integrated Circuit Technology Integration Research Platform, Deputy Dean and Professor of the School of Microelectronics Zhou Peng, Deputy Director of the Institute of Biomedical Engineering Technology, Professor of the School of Information Science and Engineering He is An, Xu Feng, director of the Key Laboratory of Electromagnetic Wave Information Science of the Ministry of Education, deputy dean and professor of the School of Information Science and Engineering, and Fang Hongbin, a young researcher of the Institute of Engineering and Applied Technology, gave reports successively.

The innovative development path of integrated circuits

"This year marks the 76th anniversary of the birth of the transistor. I would like to review with you how transistors and integrated circuits developed, and how they have changed the lifestyle of everyone here." At the beginning of the report, Liu Ming gave a brief overview The development history of integrated circuits.

Liu Ming The pictures in this article are all provided by Fudan University

"The invention of the transistor is not an isolated event." Historically, there have been countless scholars who have contributed to the field of integrated circuits. The transistor was based on the predecessors' profound understanding of solid-state physics and was driven by market demand. And developed. Liu Ming concluded: "Devices and circuit architectures with different principles and functions enable integrated circuits to be widely used in different fields. Researchers' breakthroughs in device and circuit architecture will also allow integrated circuits to have new applications." Scenes."

Liu Ming introduced that increasing integration, improving performance, and reducing power consumption are the core goals of the development of integrated circuit technology. Over a long period of time, the continuous shrinkage of transistor sizes has driven the rapid progress of integrated circuit technology. Materials, device structures, lithography technology, packaging, and even EDA tools are all constantly evolving in the process of shrinking the size of integrated circuits.

In order to further shrink the size of transistors, the manufacturing processes and equipment of integrated circuits are becoming increasingly complex. Taking EUV lithography machines as an example, Liu Ming introduced the diversification of photolithography technology from light sources, lens materials and structures, and pattern transfer modes. Global innovation, pointing out that EUV lithography machines are the most complex and most precise machines ever manufactured by humans, and are also the product of global industry cooperation.

At the same time, advanced process chips are faced with the constraints of the "area wall" of photolithography. The area of ​​a single chip (die) has approached the limit of the graphic projection capability of the photolithography machine. Liu Ming pointed out that the integrated circuit industry is entering a new stage of development, that is, a stage in which systems, design, manufacturing and packaging achieve coordinated development to cope with current challenges. Integrated chip technology - chip-level semiconductor integration technology, which can decompose complex chips into core particles and use silicon precision manufacturing technology to integrate several core particles to achieve complex system integration and applications, break through the "area wall" limitation, and realize chip Performance improvement and function expansion.

To achieve sustainable development, we need more efficient and stable new solar cells

Zhan Yiqiang pointed out in a report titled "Research on Flexible, Efficient and Stable Perovskite Solar Cells" that at present, global energy and environmental issues are prominent, and the coordinated development of perovskite cells can effectively improve the supply capacity of solar photovoltaic products and technology. Promote sustainable development and accelerate the realization of carbon peak and carbon neutrality.

Zhan Yiqiang

Utilizing the semiconductor photovoltaic effect, solar energy has become an ideal clean energy and an important way for humans to efficiently utilize solar energy. Zhan Yiqiang introduced that new perovskite photovoltaic technology is one of the key research directions of the National Key Laboratory of Photovoltaic Science and Technology at Fudan University. It has achieved high efficiency comparable to monocrystalline silicon technology, has low cost, low energy consumption, and is compatible with large-scale applications. Unique advantages such as area and flexibility.

Facing the major national needs, the team strives to solve major scientific problems in advanced solar cell technology, focusing on key tasks such as the practical cutting-edge technology of low-cost and high-efficiency perovskite cells, and creating a source of photovoltaic technology innovation.

Through more than ten years of systematic research, the team has used covalent cross-linking strategies to initially solve the core issue of intrinsic stability of perovskite, and combined with surface and interface modification, stress control and other means to collaboratively improve the working life of the device . Relevant research has developed a low-temperature preparation process suitable for flexible substrates, using down-conversion luminescent materials to broaden the light absorption range of flexible perovskite and prepare more efficient flexible perovskite photovoltaic devices. Facing the future, flexible perovskite devices are also likely to be widely used in aerospace, building photovoltaic integration, transportation, wearable and portable devices and other fields.

The development of AIGC large models is the general trend in the future

AIGC, or "generative artificial intelligence", is subverting scientific research in various fields and at the same time having an impact on human society. Yan Bo introduced in a report titled "AIGC New Engine and Its Multidisciplinary Cross-Integration Application" that AIGC includes three major elements: computing power, data, and algorithms, and its technology has evolved into three cutting-edge capabilities: twinning, editing, and creation. Using the AIGC X model can introduce new methods, discover new materials, and generate new data for scientific research.

Yanbo

"Quantitative changes have triggered qualitative changes in AIGC's generation capabilities. For example, the recently popular large language model ChatGPT has allowed us to see the results of a qualitative change." Yan Bo said that currently, the potential of AIGC models is highlighted and can achieve high-fidelity 3D construction. The model can synthesize talking figures in real time, and can also synthesize creative content based on text to generate high-resolution videos.

With the development of society, the industry has put forward new demands in the fields of smart security, smart medical care, smart terminals and other fields. In this regard, AIGC can better solve the problem of media adaptation and editing, and the inaccuracy of small objects at long distances. Problems such as low quality and low resolution cannot be seen clearly.

AIGC’s powerful editing capabilities are also playing an important role in real society. The "full concentration" surveillance video concentration system developed by his laboratory can achieve rapid positioning and recognition of conditional semantics, achieve the purpose of quickly solving crimes, and reduce search time by 99%. The Shenzhen Public Security Department has successfully applied this complete surveillance video extraction technology and system in case handling and investigation.

AIGC’s twin capabilities have also been cleverly used in the restoration of historical materials at Fudan University History Museum, which can reproduce important historical image materials that carry the memory of Fudan University in high definition.

In the application of artificial intelligence in the medical field, the laboratory has studied AIGC medical applications and successfully developed the core algorithm and hardware system of "Endoscopic Smart Eye". Relevant results have benefited more than 80,000 patients in the past four years.

Yan Bo finally concluded that the development of AIGC large models is the general trend in the future, which will strongly support AI scientific research and bring new research paradigms.

The two-dimensional semiconductor industry integrated with silicon is a new blue ocean

Integrated circuits are the basic support for the development of my country's modern industries and an important component in achieving high-level independent innovation in science and technology. The application of crystalline silicon helps transistors realize the "incremental" miracle of integrated circuits, which is a basic technology closely related to the future of mankind.

In recent years, problems and challenges in the field of integrated circuits have become more and more obvious. The most important ones are process problems caused by shrinkage, energy consumption problems caused by excessive energy density, and speed mismatch caused by storage walls. question. The performance development of integrated circuits is gradually slowing down due to limitations of the three physical principles of size shrinkage, power consumption efficiency and storage walls.

zhoupeng

In the field of integrated circuits, problems can only be overcome through innovation, and bottlenecks in new materials, new structures and new devices need to be constantly broken through. Zhou Peng's team turned its attention to two-dimensional materials with richer physical properties and more diverse properties to build new devices and find solutions to the current problems of integration density and energy consumption for silicon.

"Thanks to its unique and excellent performance, two-dimensional semiconductors are expected to overcome traditional technology bottlenecks and alleviate energy consumption challenges and storage difficulties in the development of silicon-based integrated circuits." Zhou Peng said. In recent years, Zhou Peng has led his team to achieve double surface channel transistor (TSCFET), ultra-thin multi-channel gate-around transistor (MBCFET), silicon-based two-dimensional heterogeneous integrated stacked transistor (CFET), two-dimensional PN A series of research results such as Flash memory.

"My report today is to prove that two-dimensional semiconductors are feasible." Zhou Peng said that two-dimensional semiconductors have unique electrical, thermal, chemical and optical properties, and are expected to achieve ultimate gate length scaling and can be used for To build compact back-end CMOS circuits (CFETs), laboratory-based ultra-large-scale 2D transistors have shown good performance, and pathways for industrial-scale production of 2D-FETs are also being developed.

Zhou Peng believes that the two-dimensional semiconductor industry integrated with silicon is a new blue ocean that requires further innovation-driven and continuous core technology research and development.

Challenge the "forbidden zone": Ultrasound evaluation and regulation of the skeletal system

More than twenty years ago, bones were a "forbidden area" for ultrasound diagnosis. His team challenged the "forbidden zone" and made great progress in bone ultrasound research. This time, he brought a report titled "Ultrasound Evaluation and Regulation of the Skeletal System".

Osteoporosis has become the second most common disease in the world and affects human health. There are more than 100 million osteoporosis patients in China. It can be said that skeletal ultrasound research is oriented to people’s life and health and major national strategic needs.

When diagnosing osteoporosis in adults, X-rays, CT, and dual-energy X-rays are mainly used. "These methods can reflect the 'quantity' of bones, that is, bone density; but they cannot reflect the 'quality' of bones, such as bone elasticity." He De'an pointed out.

Over the years, ultrasound has received attention and clinical application due to its many advantages. However, most of the existing bone ultrasound imagers use the ultrasonic transmission method, which has limited measurement locations and can usually only measure the calcaneus. The imaging spatial resolution is low and can only roughly display the internal morphology and contour information of the bone.

He’s safe

Faced with many challenges, the Tadian team proposed a theoretical model of ultrasonic scattering in honeycomb cancellous bone, constructed a theoretical model of ultrasonic propagation in mucus-filled, multi-layered tubular long bones, and optimized the solution of inverse problems based on meta-learning. Algorithm, a new bone ultrasound diagnostic instrument based on ultrasound backscattering method was developed.

Currently, this bone ultrasound diagnostic instrument has been used in the Dixing II experimental head-down bed rest experiment. It provides an important theoretical basis and a large amount of valuable experimental data for studying the law of bone loss in microgravity environments and combating bone loss for Chinese space station astronauts.

A revolution in radar technology: synthetic aperture radar

At the Shanghai Science and Technology Award Conference held last week, the team led by Xu Feng won the first prize of Shanghai Natural Science Award. He delivered a speech on "Microwave Vision and Radar Intelligent Target Recognition" on the podium of Xianghui Hall.

Xu Feng

During World War II, the most traditional radar was equipped with a circular display interface, which could display the detected target on the interface in the form of a cursor. What Xu Feng’s team is studying is synthetic aperture radar (SAR), which can be installed on flying platforms such as satellites or aircraft to image the earth all day and all day long, achieving a leap from one-dimensional ranging to two-dimensional high-resolution imaging. , is a revolution in radar technology.

Our country has conducted independent research and development for more than 40 years and has launched dozens of radar satellites, but it still needs to break through the key application bottleneck of radar image interpretation. "Radar satellites can obtain massive amounts of data every day, and it is very inefficient to rely solely on trained professionals for manual interpretation. Our country urgently needs to carry out basic research on automatic interpretation of SAR images," Xu Feng said.

Academician Jin Yaqiu proposed a new concept, which is to develop a physics-inspired artificial intelligence technology specifically used to automatically interpret microwave domain radar images. This is the so-called microwave vision. Xu Feng introduced the differences between microwave vision and traditional optical vision in terms of working principles, image properties, and cognitive mechanisms. If optical vision is bionic vision based on optically stimulated innate visual neural networks and acquired big data learning and training, then microwave vision is interpretable physical vision based on electromagnetic physical mechanisms and can adapt to sparse data.

In addition to discussing the physical basis, intelligence basis, and cognitive basis of microwave vision, and proposing a variety of intelligent target recognition algorithms, Xu Feng's team also designed a set of UAV-borne radar for integrated applications of detection, imaging, and recognition. Imaging algorithms and systems. In the video played at the report meeting, the drone slowly rose above the light grass and performed high-resolution SAR imaging of the main and southern districts of the Handan campus, achieving high-resolution imaging without assisted positioning. The team is currently working on Develop a microwave vision principle prototype for on-board and on-board intelligent processing, and further push the research results into actual systems.

After various large models came out, Xu Feng’s team did some tests on them and found that it was difficult to directly apply radar images with strong physical properties, indicating that it did not learn systematic physical knowledge. Therefore, he believed that physical intelligence Science for AI, represented by AI, has a lot of room for development. Faced with the question "Does ChatGPT mean the arrival of the singularity?", Xu Feng gave his own response at the end of the report, "No matter whether the singularity arrives or not, we believe in science!"

Build future robots and get inspiration from nature

Where will robots go in the future? Animals and origami provide endless inspiration for scientists.

Our common robot cases, such as service robot arms and four-legged robot dogs, have always drawn inspiration from nature. "In Fang Hongbin's view, bionics is an important force in promoting the progress of robotic technology.

The design, modeling and control of bionic mobile robots are both new hot spots and new challenges. When faced with narrow and restricted environments, such as search and rescue of survivors in disaster ruins, inspection and cleaning of industrial pipelines, detection and treatment of human gastrointestinal tract, covert reconnaissance of battlefield information, etc., traditional wheeled and legged robots are unable to do anything. There is an urgent need to further learn from animals and propose new robot structural designs.

Fang Hongbin

How to learn from animals? Fang Hongbin said: "Legless animals have the ability to move well in restricted and narrow unstructured environments. For example, earthworms have the ability to move both on the ground and in underground pipes." Learn the morphological characteristics and movement of legless animals Mechanism, the development of bionic legless mobile robots is an important direction for future robot development.

In recent years, the Bionic Structure and Robotics Laboratory of Fudan University Intelligent Robot Research Institute has achieved many innovative results in the field of worm-like mobile robot research. Fang Hongbin led the design of a variety of worm-like multi-modal mobile robots such as "squirming and swimming", "squirming and swinging", and "squirming and rolling", which can move efficiently in complex environments including pipelines, waters, and gravel fields. Fang Hongbin also innovatively introduced the "folding" idea into robot development, transforming the traditional "three-dimensional design-three-dimensional processing-final assembly" robot preparation process into "two-dimensional crease processing-folding", and designed a variety of origami-like worm-like mobile robots , constitutes an important direction for future robot development.

Fang Hongbin believes that there are both opportunities and challenges for the research of bionic robots and origami robots. In the future, robots will show more rigid-flexible coupling characteristics and reconfigurable characteristics, and the performance of robots will also improve in multi-modal motion, high performance, miniaturization and Keep moving forward in the direction of intelligence.

The above is the detailed content of Integrated circuits, AIGC, robots... 7 Fudan scientists foresee the future this way. For more information, please follow other related articles on the PHP Chinese website!