机器人产业发展的核心技术必须是自主可控的
发布时间:
24-07-09
Intelligent robots are leading a new round of global scientific and technological revolution and industrial transformation. The robot is the carrier of chip technology research, and the chip is the guarantee for the realization of the robot's function. Cultivating and promoting the development of China's intelligent robot core chip technology and industry, and proposing a feasible development route for intelligent robot autonomous and controllable special chip technology, has important industrial value and strategic significance to solve the "stuck neck" problem of robot development, promote industrial optimization and upgrading, and achieve a jump in productivity.
中国工程院院士王耀南研究团队在中国工程科学2022年《中国工程院学报》第4期发表《我国智能机器人核心芯片技术发展战略研究》一文。阐述了智能机器人核心芯片技术在推动技术自主可控、实现经济高质量发展、满足居民美好生活需要、提升国家核心竞争力等方面的重要价值。梳理了相关政策、技术和产业的国际进展，分析了我国在智能机器人核心芯片发展方面的基本优势和面临的问题。基于多架构路线和技术方案比较，论证了我国智能机器人芯片技术的发展路线，并据此提出了领域发展战略，形成了2035的重点任务和发展路线图。文章建议，应将智能机器人芯片自主可控发展上升为国家战略，明确顶层设计。设立智能机器人芯片重大科技专项，加大科研投入; 出台鼓励智能机器人芯片技术研究和产业化应用政策，带动产业链升级; 落实智能机器人芯片人才培养和发展办法，促进科技和产业的健康发展。
The picture comes from the Internet
I. Introduction
作为新一轮科技革命和产业革命的代表性成果，智能机器人在人工智能，机械工程，控制论，材料，计算机等多学科领域研发制造，产业链覆盖面广，带动力强，是推动我国工业、农业、国防等各大产业转型升级的战略性产业，也是衡量国家科技创新实力的重要标志。随着智能机器人逐渐融入人类的生产和生活，所面临的环境越来越多样化，所从事的任务越来越复杂，以及高可靠性的应用要求，高动态、强对抗对智能机器人的算法和计算能力提出了高要求，但我国在机器人所需的核心高性能芯片领域还比较薄弱，核心零部件的自主研发和生产能力与国际领先水平相差较大，而且运算能力、稳定性和集成度等性能都不好。随着国际形势日益严峻、中美贸易摩擦升级、逆全球化趋势加剧，迫切需要加强国内制造能力，掌握新兴产业主导权，实现智能机器人芯片技术及其产业链的自主可控。
目前，我国在智能机器人核心芯片领域的研究还处于起步阶段，制定合理的战略规划对其发展起着至关重要的作用。现有研究系统梳理了智能机器人和芯片技术的发展现状，提出了发展策略建议。相关研究围绕人工智能核心软硬件在技术、产业、政策等方面的发展现状，总结了我国面临的生态体系不完整、基础理论研究存在不足、创新环境不完善等问题，并对其技术和产业发展提出了思路和目标。本文针对地空协同多模态智能机器人，分析了该类机器人的发展现状，指出其缺乏系统的发展规划，没有形成全链条发展模式，软硬件结合不足，并给出了相应的发展建议。本文介绍了人工智能芯片的概念、分类和发展历程，对其产业发展现状和趋势进行了调研和辨析，提出了人工智能芯片产业发展的建议。《先进半导体材料及辅助材料》编写组提出了半导体材料及辅助材料系统开发、上下游协调发展、可持续发展的思路，形成了集成电路发展的相关对策建议。工业和信息化部、发展改革委、科技部等15个部门联合发布了《机器人产业发展 “十四五” 规划》，为我国智能机器人产业发展提供了指导。
Robot is the carrier of chip technology research, and chip is the guarantee of robot function realization, and the development of the two has a profound internal relationship, which is what the current research has failed to explore in depth. To this end, the Chinese Academy of Engineering has launched relevant research in a timely manner for the overall and key issues of the independent and controllable development of the core chip of intelligent robots, aiming to sort out strategic planning, demonstrate technical paths, and propose technical solutions. This paper analyzes the important value and international progress of intelligent robot core chips, summarizes China's basic advantages and problems, focuses on demonstrating the technological development route and proposes a development roadmap for 2035, in order to provide reference for national industrial policy, technology planning, technical research and other research.
Second, the strategic significance of the development of core chips for intelligent robots
The intelligent robot core chip is a high-performance processor specially used to realize robot computing functions such as perception, control, navigation, decision-making, and planning, and is the key to the robot to complete various tasks in complex environments. Through the research on the development strategy of the core chip technology of intelligent robots, it can provide ideas for the research of related technologies, promote the independent and controllable technology chain and industrial chain, and promote high-quality economic development.
(1) Promote the important track of independent and controllable technology
目前，智能机器人核心芯片产业正面临严重的内忧外患。内部隐忧在于机器人芯片产业链核心技术严重不足，国产化率很低，国产芯片封装测试初步可控，但设计、制造等高端环节自主可控率严重不足；外部威胁在于当前大国博弈的国际形势，西方国家对中国高端装备、材料和技术实行封锁，对先进芯片实行禁运，遏制中国发展。因此，迫切需要开展智能机器人核心芯片技术方案的论证，根据算法模型的特点设计芯片计算架构，根据计算架构优化算法模型和编译，以达到高性能的目标，低功耗、低时延，推动 “直道追赶” 与 “变道超车” 并行，积极抢占集成电路技术创新竞争高地，提升智能机器人核心芯片技术链和产业链的自主可控能力。
(2) An important support for achieving high-quality economic development
核心芯片技术的发展可以提高智能机器人的运行能力和适应性，进一步释放对智能机器人的需求，从而带动整个产业链的技术升级和产品结构优化，对振兴实体经济作用显著，进而成为经济由虚转实的重要手段。大力发展智能机器人及其核心芯片技术，可以带动制造业整体提升，成为制造强国建设的突破口。此外，作为信息革命的代表性产业，智能机器人核心芯片产业的发展，可以进一步融合大数据、云计算、物联网、5g通信等新兴技术，构建新的产业体系，成为推动中国经济高质量发展的 “火车头”。
(3) An important guarantee to meet the people's needs for a better life
我国社会主要矛盾已经转化为人民日益增长的美好生活需要和不平衡不充分的发展之间的矛盾。智能机器人核心芯片技术的发展，可以提高智能机器人的应用范围，大幅提升社会生产力，使人从重复的机械劳动中解脱出来，创造更多水平和质量更高的就业岗位; 智能机器人适应性的提高，还可以增加特种作业的需求，让人们远离危险的工作任务，在消防、治安、搜救、排爆等工作中发挥重要作用; 智能机器人的大规模使用，可以有效提高生产效率，从而减少生产过程中的消耗和污染。推动智能机器人产业发展，可以有效节约资源、减少排放，成为 “两型社会” 建设的重要支撑。
(4) An important carrier to enhance the core competitiveness of the country
大力推进智能机器人核心芯片技术和产业发展，有助于保障中国国防安全，抢占全球科技制高点，从而有效提升国家核心竞争力。目前，无人作战已逐渐成为现代战争的主流，无人系统作为一种具有自主机动能力的智能机器人，其在态势感知、规划决策、多机协作、运动控制、等都离不开智能机器人核心芯片的计算和处理能力，本项目的研究将有助于推动军事智能化的发展，推动我国国防现代化的进程。作为新一轮科技革命的典型产物，机器人核心芯片是衡量国家科技创新实力的重要标志，而大力发展机器人核心芯片，可以快速推动我国相关高新技术产业的发展，提升国家的核心竞争力。
3. The international development status of core chips for intelligent robots
(1) Policy planning
In recent years, the global chip shortage has had a serious impact on all intelligent industries, and the world's major developed countries and regions have formulated a series of policy plans for chip development, focusing on technological innovation, interface standards, financial plans and other comprehensive layouts to seize the leading position in chip technology development.
In 2020, the United States introduced the "American Foundry Act of 2020", which recommended that Congress tighten the bottleneck of chip manufacturing technology to prevent China from overtaking the United States in the next few years. Subsequently, the American Innovation and Competition Act of 2021 (USICA) was introduced, which advocated tough countermeasures against China to improve the ability of the United States to compete with China in science and technology. In 2020, the European Commission issued the "Joint Statement on the European Processor and Semiconductor Technology Plan", which is expected to invest 145 billion euros to help the development of the semiconductor industry. In 2020, Japan released its "Green Growth Strategy", focusing on the semiconductor industry as a key development area. South Korea announced in 2020 that it would invest USD 9.5 million over the next five years to develop materials technology for sub-5 nm semiconductor lithography processes.
The traditional chip powers intend to rely on their rich experience, strong innovation capabilities, and strong economic strength, under the guidance of policies and regulations, and with the national strategy to reach a consensus among all parties, and guide superior resources to the chip field, so as to lead a new round of scientific and technological revolution and further strengthen their competitive advantages.
(2) Technical aspects
In the new wave of scientific and technological revolution, the application scenarios and operation requirements of robots are becoming increasingly complex, and higher requirements are put forward for the algorithm computing power of environmental perception, decision-making planning, navigation and positioning, and motion control. At present, the computing system based on general-purpose embedded processors and programmable logic controllers cannot meet the needs of the perception, cognition and sensitive movement ability of future robots. Therefore, it is urgent to develop the core chip of the robot under the new situation, integrate the latest artificial intelligence algorithms, and improve the intelligent autonomy of the robot.
In terms of robot intelligent algorithms, the empowerment of artificial intelligence has brought about rapid development:
(1) Artificial intelligence improves the efficiency of environmental perception. For object recognition, a series of classic frameworks such as SSD, YOLO, R-CNN are proposed, so that intelligent robots can simulate the computing mode of the human brain and rely on multi-layer convolutional neural networks to greatly enhance the ability to perceive the environment.
(2) Artificial intelligence improves planning and decision-making capabilities. The rapid development of artificial intelligence technology has greatly improved the autonomy of intelligent robots in planning and decision-making, and they can still carry out normal tasks in highly dynamic, highly complex, and strong confrontational environments.
(3) Artificial intelligence strengthens multi-machine collaboration capabilities. Based on "mobile energy and information interconnection" and focusing on the core of "model algorithm and data computing", intelligent robots can realize autonomous operation, man-machine collaboration, and machine-machine collaboration, which will become the main way of intelligent robot application in the future. Artificial intelligence technology provides strong technical support in interoperability, autonomous decision-making, and cluster control algorithms.
In terms of intelligent robot computing hardware, it is facing the problem of Moore's Law failure, and is gradually developing in multiple directions such as high energy efficiency, specialization, and customization.
(1) In the general artificial intelligence chip architecture, it is gradually evolving to brain-like and neuromorphic. In 2019, Tsinghua University released the world's first heterogeneous fusion brain-like chip - "Tianji", which allows bicycles to truly realize automatic driving and appear on the cover of Nature;
(2) In terms of manufacturing process, chip manufacturing has entered the atomic age. In 2021, Hunan University and Samsung Advanced Institute of Technology in South Korea published a review article in Nature, which discussed the key material parameters of 2D semiconductors and traditional 3D semiconductors, and summarized the ways to push the limits of 2D transistors.
(3) In terms of chip design, artificial intelligence (AI) automatic design may be realized in the future. In 2021, the Google Brain team and Stanford University published a paper in Nature, in which the AI-based chip design method was improved and applied to the next-generation tensor processing unit (TPU v4) accelerator officially released at the Google I/O 2021 conference.
在机器人智能算法与芯片硬件的融合方面，类脑芯片是智能算法的物化解决方案，具有天然节能的优势。众所周知，算法、算力和数据是人工智能技术发展的三驾马车，人工智能技术的引入提升了机器人的算法效率，各种传感器为机器人引入了海量数据，而类脑芯片的出现，成为算法和数据的载体。智能机器人芯片的信息处理单元架构逐渐向仿生脑结构方向转变，“存储与计算一体化” 的类脑芯片成为人工智能技术创新的基础，并且在运行一些智能算法时具有高能效的优势。底层的非易失性核心器件也在逐步取代易失性核心器件，从而更好地实现非结构化环境和复杂任务的智能计算功能。
上一个:
下一个:
相关新闻
2025-04-30
2025-04-20
2025-04-10
2025-03-31
