The past year has been an extraordinary year. Digital technology has penetrated into all aspects of the economy and society as never before. We are in a new technological innovation cycle. This point seems to have become more widely felt as hard technology innovation and industry-university-research transformation have become louder and louder.
If we take the year as a unit, which cutting-edge technologies and innovative breakthroughs are coming from behind the scenes to the stage, from the laboratory to the industry, and will soon affect each of us...
In order to allow more people to accurately grasp the cutting-edge technology trends, and to help more people connect the internal clues of technological breakthroughs and industrial trends, and to help everyone see the technology-driven future in advance. Next, let us learn about the top ten cutting-edge scientific and technological developments in this extraordinary year of 2022.
High-performance computing - High-performance computing moves towards the "CPU+GPU+QPU" era.
computing power is a new productive force in the digital economy era, and is widely integrated into all aspects of social production and life. High-performance computing is the representative of advanced computing power and , and is also "the most important weapon of the country." The development and application of high-performance computing cutting-edge technology leads the development trend of the entire computing field and even triggers epoch-making leaps, which deserves the continued attention of the industry and society.
In recent years, various artificial intelligence applications such as AI large models, AIGC, autonomous driving, protein structure prediction, etc. have emerged in large numbers, which have had a significant impact on the development of high-performance computing. 2022 will be a year of accumulation for the development of high-performance computing technology. The iteration and accumulation of high-performance computing in terms of architecture, hardware, and software will accelerate the intergenerational transition to 2.0 and enter the new era of 3.0 under the continuous shaping of "application drive".
Ubiquitous operating system - Ubiquitous operating system accelerates the comprehensive integration of humans, machines and things
The operating system is the most critical layer of system software in the computer system and is the core of the computing system.
The ubiquitous computing era in which human society, information space, and the physical world are deeply integrated is beginning. New scenarios that integrate massive, heterogeneous, and heterogeneous resources of "humans, machines, and things" are emerging, and the complexity of the resources required to be managed is increasing exponentially.
It has become a development trend to build a ubiquitous operating system that manages various ubiquitous facilities/resources and supports digital and intelligent applications in various scenarios.
AIGC - Comprehensive virtual humans appear in the AIGC field
With the development of deep learning, AI-generated virtual content AIGC is penetrating into various fields such as images, videos, CG, AI training data, etc., and even covers multi-modal virtual human technology at the same time.
Virtual digital human refers to a comprehensive product that exists in the non-physical world, is created and used by computer means such as graphics rendering, motion capture, and speech synthesis, and has multiple human characteristics. It is currently divided into two categories: "CG modeling + real-person driver" and "deep synthesis + computing driver".
Among them, the final effect of computing-driven virtual humans is affected by a variety of AI generation technologies, such as speech generation, text generation and understanding, image generation, etc.
Content creation has gradually shifted from being highly dependent on people in the early days to "manpower + computing power". In addition to being directly applied to content-related business scenarios (news, audiobooks, industrial design, etc.), AI also greatly lowers the threshold for generation and promotes highly customized, automated, and democratized content creation.
robot - flexible material innovation promotes robot bionic refinement
Tactile perception is the current focus of research in the field of robot perception completion.As one of the five forms of perception, touch plays an irreplaceable role in key links such as robots understanding the working environment (pressure, sliding, contact, etc.), acquiring the characteristics of contact objects (form, texture, smoothness, temperature and humidity, etc.) and interacting with them.
The tactile sensing system is divided into sensors and actuators. Benefiting from the breakthrough progress of flexible materials, tactile sensing technology has repeatedly released representative results in the scientific research community, and has been developed and tested in fields such as robot hands, tactile gloves, health detection equipment, and smart cockpits. tactile actuators are more complex, and the personal wearable device scene is further away from the market. Force feedback gloves using mechanical braking have entered the era of personal consumer goods; in addition, tactile gloves developed using microfluidic technology have been unveiled at the CESth annual exhibition.
In the next 1-3 years, with the advancement of flexible electronics technology and machine learning algorithms, the spatial resolution and accuracy of tactile sensors are expected to be significantly improved. It is of great significance to improve the dexterous operation level of robots and bring about an upgrading of human-computer interaction experience, thereby further expanding the application space and service level of intelligent robots.
includes industrial robot teleoperation, medical surgical robots, simulation training, aerospace exploration, smart cockpits and other scenarios that will more complement the tactile perception and expand to the next generation of virtual reality interactive terminals, terminal prosthetics and other scenarios.
Metaverse - XR creates a second world and matures the Metaverse
Just this year, the overall technology ecology of the Metaverse has matured. By improving optical devices, spatial computing, heterogeneous computing system, rendering engine, interactive freedom, positioning methods and other factors, various problems such as discomfort in viewing and rough images in the past have been solved.
In the past, we relied on audio recording, video recording and other forms to span time, and after using mobile phones, the Internet, etc. to span 2D space, XR has taken us to achieve further leaps.
On the one hand, XR helps us transcend the 3D space and break through the limitations of on-site observation and operation in a more three-dimensional and realistic way. The cost of information restoration and transmission is further reduced; on the other hand, XR helps us transcend the limitations of reality and allows us to communicate and entertain in the second space.
Quantum computing - Quantum computing revolutionizes the classic computing paradigm
The Chinese team, headed by University of Science and Technology of China , has reached the forefront of the world in quantum computing hardware research and development.
Currently, our country is the only country in the world that has reached the milestone of "quantum computing superiority" in two physical systems. specifically refers to the photon route represented by Jiuzhang and the superconducting route represented by Zu Chongzhihao . In terms of applications and supporting facilities, finance, medicine, automobiles, chemistry and other fields have clarified the use of quantum computing for specific problems.
Under specific problems such as microscopic simulation and complex modeling, quantum computing has demonstrated advantages that are difficult to achieve by classical computing. In the future, quantum-classical hybrid architectures may appear in supercomputing centers, where quantum computing and classical computing will cooperate to solve specific large-scale problems.
Brain-computer interface - Invasive brain-computer interface is implemented in difficult medical scenarios
Nowadays, with the development of wireless communication , multi-channel flexible electrodes, implantation methods, chips, machine learning algorithms and other technologies, invasive brain-computer interface is gradually overcoming engineering and clinical difficulties.
This year, represented by Nature’s thought typing technology, the invasive brain-computer interface has shown satisfactory clinical trials and commercial development has begun to take shape.
Under the guidance of national policies, the development of invasive brain-computer interfaces in my country has also begun to accelerate: Tsinghua University Li Luming's team developed a second-generation brain pacemaker; Ruijin Hospital launched a clinical trial for the treatment of severe depression; Zhejiang University and the Neurosurgery Department of the Second Hospital of Zhejiang University The department has completed the first clinical trial of invasive brain-computer interface in China, installing a mechanical arm for an elderly person with high paraplegia; and universities such as Tsinghua University, Tianjin University, Shanghai Jiao Tong University, Chinese Academy of Sciences, and South China University of Technology have established key scientific research teams.
After scientists further understand how the brain works (such as sensory areas), brain-computer interfaces will play more roles in helping patients restore specific sensory abilities such as touch and vision.
AI chip - new AI chip leads the post-Moore era
As AI is widely implemented in various scenarios, AI chips that traditionally rely on process technology and improvements are difficult to meet demand.
Among the three future evolution routes of integrated circuit , "More than Moore" represented by complete architectural innovation has become the key direction of the next generation of AI chips. Among , brain-inspired computing, integrated storage and calculation, quantum computing, and data flow AI computing are all options.
Take the event-driven neuromorphic chip developed after the human brain as an example. Because it imitates the way electrical impulses are transmitted between neurons as much as possible, the neuromorphic chip naturally conforms to the event-driven mechanism, integrates storage and calculation, and significantly reduces latency and energy consumption.
Gene editing - CRISPR helps make gene editing controllable and reliable
Gene editing technology represented by CRISPR-Cas9 is maturing step by step, moving from the laboratory to clinical application.
According to the industry, because cells (especially allogeneic cells) needed by patients can be quickly and targetedly prepared based on the patient's condition, it is expected to promote personalized treatment of major diseases in the future and change the past medical process of standardized production and distribution of drugs. The application in synthetic biology can use different gene control modules to create more complex biological systems.
Molecular breeding , as one of the representative fields, compared with traditional screening methods using phenotypes and natural selection, combined with gene editing, can purposefully change the species' stress tolerance, composition, yield, reproduction and other traits, shorten the species domestication cycle, and create species with better traits. In addition, it is also playing a role in assisting other medical methods, DNA storage and other fields.
A brief summary: Based on gene editing technology, life science research is expected to achieve "precise planning + fine transformation".
Multi-modal learning - multi-modal and multi-dimensional large models predict general intelligence
Due to their strong versatility and few-sample learning capabilities, large models are bringing intensive new development models and business models to AI. At the same time, the emergence of cross-modal pre-training models (such as DALL·E, CLIP) heralds the achievability of general intelligence. It is generally believed in the industry that "develop once and use it for life".
Multi-modal learning has become an important trend. It can be applied in normalization, transformation, translation alignment, fusion and collaborative learning. According to the downstream tasks, can be divided into comprehension tasks such as visual question and answer, visual reasoning, image and text retrieval, and generative tasks (text to generate images).
Due to its cross-domain versatility, large-scale pre-trained models may play a role as a middle layer in a similar infrastructure ecosystem in the future, taking on a transitional role for different industry ecosystems.
At the application layer, it also brings opportunities for the application and development of artificial intelligence in various industries, such as automated content generation, content translation, robot dialogue, etc.Large models also improve their performance in this process and exert the iterative effect of data closed loop.
Conclusion
my country takes technological innovation as the first driving force for development, deeply implements the "innovation-driven development" strategy, and embarks on a new journey to build a world power in science and technology. Key core technology patents are important tools for the country and are of great significance in promoting economic development and technological progress.
The development of technology has no end. It is the eternal trend to integrate technology into the real economy to promote its high-quality development , making life more convenient and making society better.
Editor in charge: Sun Lang
Producer: Li Hongmei
Article reference:
1. "Top Ten Technology Application Trends in 2022" Tencent Research Institute
2. "Baidu releases the "Top Ten Technology Frontier Inventions" for the first time Achieving multiple international cutting-edge technological breakthroughs" Crack quantum encryption, fastest matrix multiplication , etc. are on the list" Industry and Information Headlines
