Editor’s Note: Artificial intelligence has become the core driving force reshaping the global landscape and leading a new round of technological revolution and industrial transformation. The integration and innovation of artificial intelligence and robotics lay the technological foundation for the arrival of the smart economy era. At the national strategic level, there is a high emphasis on the coordinated development of artificial intelligence and robotics, incorporating embodied intelligence and intelligent robots into future industries. To seize development opportunities and build an innovative ecosystem for the smart economy, the Jiliang Research Institute has systematically sorted out the development trends of the artificial intelligence and robotics industries, as well as the experiences and innovative measures from advanced cities at home and abroad, proposing recommendations for promoting innovation through integration.1. Overview of the Artificial Intelligence and Robotics Industry(1) New Connotations and CharacteristicsThe technological positioning of artificial intelligence has been upgraded, becoming the underlying technological architecture that empowers various industries.Artificial intelligence is deeply integrating with digital technologies such as big data, cloud computing, and the Internet of Things, as well as industrial technologies in manufacturing, biology, and materials, becoming a strategic pivot for empowering industrial upgrades and reshaping industrial patterns. This integration promotes artificial intelligence from a single tool to a systematic engine that enhances overall factor productivity, marking the entry of AI technology development into a new stage of “fundamental innovation – industrial integration – comprehensive empowerment.”
The deep integration of artificial intelligence and robotics has given birth to new business forms and market opportunities.Through rapid iteration and deep integration, artificial intelligence and robotics technologies have completely broken through traditional application boundaries, upgrading from “manufacturing tools” to a comprehensive carrier of “intelligent terminals + system integration + service ecosystems.” Through the deep coupling of “algorithms – hardware – scenarios,” new tracks of embodied intelligence such as humanoid robots have emerged, opening up entirely new incremental markets in industrial manufacturing, logistics, consumer services, healthcare, and scientific education. Embodied intelligence, as the core intelligent form of robots, learns through interaction with the environment via a physical body; intelligent robots serve as its physical carriers; humanoid robots are a bionic branch that adapts to human environments through human-like configurations. The three form a collaborative and mutually reinforcing system of “intelligent driving carriers, carrier adaptation forms, and forms carrying intelligence.”
The artificial intelligence industry has formed a complete development chain covering basic support, core products, and application scenarios.The foundational layer integrates computing hardware such as AI chips and intelligent servers, as well as data resource platforms, providing core computing power and data support for the upper layers; the product layer encapsulates technological capabilities into process products and terminal products, achieving standardized packaging and delivery of technological capabilities; the application layer deeply integrates with vertical fields such as industrial manufacturing, healthcare, intelligent transportation, and consumer terminal products, realizing the large-scale value landing of technology. Hardware innovation and data governance drive product upgrades, algorithm optimization lowers application thresholds, and scenario feedback promotes continuous iteration of underlying architectures and product forms, forming a cross-layer mutually reinforcing and dynamically developing artificial intelligence industry ecosystem.
(2) New Requirements and DirectionsChina places great importance on the development of the artificial intelligence and robotics industries. The “State Council’s Opinions on Deepening the Implementation of the ‘Artificial Intelligence+’ Action” and the “Three-Year Action Plan for ‘Data Elements ×’ (2024-2026)” and the “Guiding Opinions on the Innovative Development of Humanoid Robots” clearly outline three major policy directions:First, focus on the landing of application scenarios,implement the “Artificial Intelligence+” action, promote large models empowering intelligent terminals and equipment manufacturing, and position embodied intelligence and intelligent robots as key future industries, promoting the integration of AI with new energy, elderly care, healthcare, and other fields.Second, focus on data ecosystem construction,relying on national-level innovation centers to formulate standards and data specifications for embodied intelligence, promote the unification of multimodal data, and build high-quality data sets in transportation, healthcare, manufacturing, and other areas.Third, focus on breakthroughs in core technologies,strengthen key technological breakthroughs in embodied intelligence, multimodal large models, and cloud-edge-end collaborative computing power, promote the research and development of industry large models, the establishment of open-source systems, and innovation tasks in a “reveal the list and take the lead” manner.
(3) New Development Trends
Models, data, and computing power are accelerating their evolution from “general support” to “vertical empowerment.” As the core contradiction of industrial intelligence upgrades shifts from technological availability to scenario adaptability, foundational resources such as models, data, and computing power are deeply sinking into vertical fields. At the model level, in the next 2-5 years, the focus will be on building a unified end-to-end large model architecture, promoting the transformation of general AI capabilities into a collaborative system of “industry foundational models + scenario lightweight modules.” Through the integration of embodied intelligence models with environmental perception, multi-step planning, and spatial motion computation, a closed loop from cognition to control will be achieved. At the data level, AI is driving a leap from “device-driven” to “data-driven,” with the deepening of the “data elements ×” action, data collection and labeling increasingly focusing on core industry processes, accelerating the iteration of general data to high-quality data sets deeply bound to industries. At the computing power level, it is transforming from resource supply to capability conversion, forming a layout of “scale nodes + hotspot optimization + edge-end adaptation,” with the heterogeneous integration of general computing, intelligent computing, supercomputing, and quantum computing becoming an important feature. By 2025, inference computing power will first exceed training computing power, driving the popularization of a “central training-edge inference” collaborative system.
Core products in software and hardware and terminals are upgrading from “single-point breakthroughs” to “batch landings.” Driven by the threefold forces of technology open-sourcing, industrial chain collaboration, and scenario openness, a spiral ascent channel of “underlying breakthroughs – mid-end mass production – terminal explosion” is formed. Technology open-sourcing significantly lowers development thresholds, promoting the universalization of underlying technologies. By 2025, over 100,000 robot applications based on open-source platforms will be developed globally, giving rise to low-cost, high-performance AI chips and sensor modules. Industrial chain collaboration accelerates product standardization and scaling, giving rise to standardized products such as collaborative robots and intelligent connected vehicles around key products like AI chips and intelligent sensors, with global sales of collaborative robots exceeding 500,000 units by 2025. Scenario openness drives the emergence of blockbuster applications through demand traction, with over 150 robots demonstrating at the World Artificial Intelligence Conference in 2025, marking the industry’s formal entry into a period of commercial explosion with multiple categories in parallel.
Key field vertical applications are expanding from “local pilots” to “full-domain penetration.” In 2024, the market size of China’s artificial intelligence industry will reach 747 billion yuan, with significant penetration rates in fields such as the Internet (89%), telecommunications (68%), government (65%), finance (64%), and manufacturing (47%). Technological penetration rates continue to rise, with collaborative robots and traditional industrial robots forming stable complements in automotive manufacturing and electronic assembly; both logistics and consumer services are in a rapid expansion phase; and high-threshold fields such as healthcare show breakthrough prospects. The scope of empowerment extends across the entire chain, breaking through the limitations of a single production link, penetrating into the entire chain of research and development design, production manufacturing, and supply chain management, promoting collaborative innovation in the industrial chain through data closed loops. Promoting the digital and intelligent upgrade of cities, expanding from local areas such as urban infrastructure and public services to complex systems such as urban lifelines, forming a multi-governance ecosystem of “government-led – enterprise collaboration – public participation,” promoting cities from “functional superposition” to “organic living bodies.”
2. Development Trends of China’s Artificial Intelligence and Robotics Industry(1) National Key Regional LayoutChina’s artificial intelligence and robotics industry has formed a “three-pole dominant” agglomeration pattern,with the three core areas of Beijing-Tianjin-Hebei, the Pearl River Delta, and the Yangtze River Delta accounting for 89.6% of the country’s AI enterprises and 84.3% of the robotics industry chain enterprises, forming the core industrial bearing area.The Beijing-Tianjin-Hebei region has advantages in basic research and policy resources,with Beijing leading in cutting-edge fields such as AI foundational algorithms and brain-like computing, holding a monopoly advantage in basic research.The Pearl River Delta region has advantages in hardware manufacturing and computing power infrastructure,with Guangdong Province currently being the only region in the country that clearly positions “artificial intelligence and robotics industry” as a unified strategic direction, holding an absolute advantage in the full industrial chain layout in intelligent hardware and autonomous driving terminals.The Yangtze River Delta has formed a full-chain layout of “basic research – high-end manufacturing – scenario applications,”with the artificial intelligence industry scale of the three provinces and one city in the Yangtze River Delta accounting for 45% of the country in 2024, possessing global competitiveness in natural language processing, machine vision, and intelligent terminals. In terms of platform layout, Shanghai (Zhangjiang Science City) is the first national pilot area for AI innovation applications, Jiangsu (Suzhou) focuses on industrial robots and core components, and Zhejiang (Hangzhou) is making efforts in humanoid robots and open-source ecosystems, forming a number of regionally leading platforms nationwide.(2) Key Enterprise Layout Paths
The artificial intelligence and robotics industry, as a technology-intensive and innovation-driven emerging industry, has key enterprise development layouts concentrated in three core path directions: AI-native innovative, scenario-deepening, and ecosystem-building, collectively forming the core driving force for industrial growth. AI-native innovative enterprises achieve underlying breakthroughs driven by technology, building competitive barriers through self-developed algorithms, data closed loops, and open-source ecosystems. They typically have high R&D investment and agile iteration capabilities, characterized by technological barriers, self-developing underlying architectures such as operating system-level models, industry knowledge base training + user feedback iteration data closed loops, and open-source core models/toolchains, attracting developers to co-build scenario applications. Typical enterprises include DeepSeek, SenseTime, Huawei’s Pangu large model, and Tencent’s Hunyuan large model. Scenario-deepening enterprises rely on the accumulation of professional capabilities in vertical fields, deconstructing industry pain points to achieve process reengineering. They deeply bind to vertical fields, accumulating industry and data assets, upgrading from single-point tools to full-process empowerment, customizing technologies for specific industry needs, and upgrading AI and robotics technologies from “tool replacement” to “value reconstruction,” forming hard-to-replicate vertical competitiveness. Typical enterprises include Yunji Technology (hotel/healthcare scenarios), Aoshark Intelligent (industrial manufacturing scenarios), and Inspur Digital Enterprises (chemical/mining scenarios). Ecosystem-building enterprises build industrial ecosystems through open platform models, focusing on “open technology base + resource integration capabilities,” expanding AI and robotics technologies from vertical scenarios to full-domain collaboration, forming a “technology – industry – business” symbiotic closed loop. Typical enterprises include Huawei’s HarmonyOS + Ascend, connecting over 700 million devices with HarmonyOS as the unified operating system, and Ascend chips providing computing power support, attracting over 300 partners to join, building an “end-edge-cloud” collaborative ecosystem.
(3) Policy Support DirectionsPolicies supporting the artificial intelligence and robotics industry exhibit specific characteristics.In terms of support methods, many policies adopt a comprehensive support approach, emphasizing the cultivation of the artificial intelligence and robotics industry ecosystem.61% of policies adopt a comprehensive support approach, supporting the entire ecosystem from factors such as computing power and data, technology applications, technological innovation, enterprise cultivation, and industrial environment. 39% of policies focus on key areas or critical links in regional industrial development.In terms of support directions, the most concentrated areas of policy support are computing power, application demonstration, technological innovation, and large models.Most regional policies provide financial support for computing power, artificial intelligence application demonstrations, artificial intelligence technological innovations, and large models; about half of the regional policies support the construction of high-quality data sets and public technology service platforms for artificial intelligence; and about half of the regional policies focus on the industrial development environment, providing inclusive policy transfers or supportive descriptions for talent, finance, and event activities.In terms of support intensity, cities such as Shenzhen, Hangzhou, and Nanjing have relatively strong support.Shenzhen has strong support in enterprise cultivation, platform services, and scientific and technological innovation activities, while Nanjing and Zhengzhou have strong support in computing power support and talent attraction, and Hangzhou has strong support in data elements and model ecosystems.3. Development Experiences of Typical Cities at Home and Abroad(1) Boston: Building a Scenario-Based Innovation Ecosystem Driven by Academic Originality and Capital Collaboration
As a global source of humanoid robot technology, Boston relies on original research from top academic institutions, promoting technology transfer through deep collaboration between capital and industry, ultimately focusing on industrial scenarios to achieve a commercial closed loop. The core of this path lies in the close linkage between academia, capital, and scene demand, rather than merely technological breakthroughs. First, academic research lays the technological foundation for the industry. Relying on over 60 universities, including MIT, continuously outputs underlying technological breakthroughs, with direct links between academic institutions and industry, ensuring that research directions align with industrial needs, while also delivering a large number of interdisciplinary professionals, with local universities training thousands of graduates in the field of robotics each year. Second, industrial capital and national strategies drive conversion efficiency. The capital side relies on donation funds from universities like Harvard and modern enterprises to provide “patient capital,” tolerating long R&D cycles. The U.S. Department of Defense continuously funds military robot research through agencies like DARPA, forming a foundation for technological accumulation. By 2025, companies like Tesla and Boston Dynamics will jointly promote the formulation of the U.S. National Robotics Strategy. Third, industrial scenario validation becomes a key verification link. Boston Dynamics’ early hydraulic Atlas, which cost over a million dollars per unit, could not be mass-produced, ultimately shifting to the electric-driven logistics robot Stretch, optimized for truck unloading scenarios, achieving sorting efficiency twice that of humans in DHL warehouses. The technological path must obey the economic viability of the scenario, reducing hardware costs through electrification and compressing training cycles from months to weeks through reinforcement learning, ultimately completing the commercial closed loop in industrial scenarios.
(2) Tokyo: A Demand-Driven Approach with Government Precision in Facilitating Technology Landing
In the development process of the artificial intelligence and robotics industry, Tokyo has built a precise support model driven by social demand, guided by policies, and cross-departmental collaboration. The key to the Tokyo model is the government acting as a “connector,” efficiently matching fragmented demands, technologies, and capital, promoting technology from the laboratory to large-scale application scenarios, ultimately forming a sustainable ecosystem defined by demand-driven R&D, policy risk mitigation, and scenario validation value. First, strengthen top-level design, the government has specifically established the “Artificial Intelligence Strategy Committee” to promote the development of Tokyo’s artificial intelligence industry. In terms of application quality, Tokyo tends to focus on the development of autonomous driving and robotics. In terms of integration quality, the government has not only led the establishment of multiple artificial intelligence research institutions, including the Artificial Intelligence Research Center and Advanced Intelligent Projects Center, but also over 20 universities, including the University of Tokyo, Osaka University, and Waseda University, have established artificial intelligence majors, laying a solid foundation for the development of the artificial intelligence industry. Second, in terms of policy support, Tokyo provides low-cost land, shared experimental equipment, and equipment subsidies to reduce trial and error costs for small and medium-sized enterprises; at the same time, it accelerates technology commercialization through tax reductions such as R&D expense deductions and industry chain docking measures. Third, the government leads the establishment of physical collaboration platforms, such as the Tokyo Metropolitan Industrial Technology Research Center, which selects real demands from enterprises, hospitals, and other frontline scenarios through an open recruitment mechanism, including aging care, transportation service gaps, etc., integrating technical resources from universities like Waseda University and enterprise engineering capabilities for joint tackling. Tokyo actively hosts international artificial intelligence exhibitions like AIEXPO, gathering industry leaders including Alibaba, Salesforce, and FujiSoft.
(3) Beijing: Using Basic Research as an Engine to Systematically Strengthen Core Technological Breakthroughs
Beijing relies on top research institutions for systematic breakthroughs in foundational algorithms and brain-like computing, building a globally leading technological barrier, establishing its monopolistic advantage in basic research. Through generational innovations in foundational algorithm architecture, brain-like hardware, and cognitive neuroscience, Beijing has formed a complete system from original innovation to technology landing. First, lead cutting-edge basic research, with the number of large model filings leading the country. Relying on 21 national key laboratories, new R&D institutions like Zhiyuan Institute and Tongyuan Institute, and universities like Tsinghua and Peking University, it gathers over 40% of the country’s top AI talents, achieving original breakthroughs in large models like Wudao 2.0 and GLM-Zero, as well as brain-like computing directions such as optical training chips and wafer-level 3D stacking chips, with 132 large model filings accounting for nearly 40% of the country. Second, explore non-consensus directions such as optical computing and non-Transformer architectures. Tsinghua University developed the “Taiji-II” optical chip, filling the gap in optical computing for large-scale neural network training; RockAI launched the non-Transformer large model “Yan,” opening up new paths for edge-side inference; the Institute of Automation of the Chinese Academy of Sciences released the Zhidong Taichu 2.0 model with hundreds of billions of parameters, overcoming challenges in multimodal cognition and encoding/decoding, building a self-developed algorithm system. Third, layout high-energy innovation platforms, focusing on common technology breakthroughs. Beijing Yizhuang established the National Embodied Intelligent Robot Innovation Center, collaborating with enterprises to overcome challenges in operational control stability and cross-entity generalization operations; it released general platforms like “Tiangong” and “Huisi Kaiwu,” achieving breakthroughs in complex motion control (speed 12 km/h) and multi-task generalized execution (accuracy 98%), with a domestic substitution rate exceeding 60%, forming a closed-loop ecosystem driven by underlying innovation to application landing.
(4) Shenzhen: Core of Soft-Hard Collaboration, Full-Stack Ecosystem Driving Product Innovation
Shenzhen has gathered over 2,600 artificial intelligence enterprises and 34 publicly listed robotics companies, forming a complete industrial chain from R&D to manufacturing, algorithms to scenarios, with the scale of the robotics industry, number of enterprises, and investment activity ranking first in the country. Its development of the artificial intelligence and robotics industry focuses on “soft-hard collaboration and scenario penetration,” constructing a product model of “agile manufacturing + scenario validation.” First, integrate technology across the entire chain, building a solid foundation for soft-hard collaboration. Relying on a strong electronic information industry cluster and the integration ecosystem of “large models + intelligent terminals,” it integrates chip design, algorithm open-sourcing, precision manufacturing, and system integration, forming a vertical system from underlying hardware to terminal applications. The localization rate of core components for humanoid robots exceeds 90%, with a local rate of 60%, achieving rapid iteration from R&D to mass production. Second, a scenario-based product matrix promotes technology commercialization. R&D is defined by demand, laying out industrial manufacturing, urban governance, and consumer electronics fields. For example, the WalkerS industrial robot supports complex scenarios such as automotive welding and 3C assembly; AI sanitation robots achieve autonomous operations in complex terrains; Huawei and Honor launched AI terminals equipped with edge-side large models, enabling multi-modal interactions. Third, innovative policy tools accelerate the construction of the industrial ecosystem. On the supply side, a hundred billion-level industrial fund is established, launching “training vouchers” and “model vouchers” to reduce R&D costs; on the demand side, four batches of “city + AI” application lists are released, promoting “first purchase and first use” policies, with state-owned orders in 2024 driving enterprise valuations to grow by 2.7 times; building the “Pengcheng Brain” open-source large model community, opening over a thousand algorithm interfaces, creating an industry ecosystem resonating with innovation and application.
(5) Shanghai: Core of Market-Oriented Data Element Allocation, Infrastructure Construction Driving Intelligent Industry Ecosystem
Since 2017, Shanghai has taken the lead in laying out the artificial intelligence industry, issuing the country’s first provincial-level AI regulation, the “Shanghai Regulations on Promoting the Development of the Artificial Intelligence Industry,” and establishing Asia’s largest intelligent computing center cluster, with over 250,000 AI practitioners, accounting for one-third of the country, forming a full-chain ecosystem covering large models, embodied intelligence, and industrial robots, with a deployment density of industrial robots reaching 383 units per 10,000 people, leading internationally. Its core experience lies in market-oriented allocation of data elements, constructing an ecosystem of artificial intelligence and robotics infrastructure. First, promote data trading and value realization. Relying on the Shanghai Data Exchange, it establishes a corpus data trading section, promoting data standardization and assetization, pioneering the “data assets on the books” model, helping enterprises realize data appreciation of over 5 billion yuan by 2024, with 12,000 data products listed and transaction volume exceeding 18 billion yuan; implementing the “Corpus Foundation Plan,” building the “Ten Thousand Volumes Silk Road” multilingual corpus database to support AI cooperation along the Belt and Road. Second, build heterogeneous training scenarios and high-quality data sets. Relying on the national humanoid robot innovation center, it creates the country’s first heterogeneous robot training ground, deploying multi-enterprise training in the same venue, collecting over 500 trajectory data points daily per machine, planning to form a tens of millions-level heterogeneous data set to alleviate AI hallucination issues; Zhiyuan Robotics builds a 2,000 square meter data collection factory, generating tens of thousands of data points daily, forming an open-source data set AgiBotWorld to support embodied intelligence model training. Third, government-industry-academia collaboration builds a multimodal data collaboration platform. The Shanghai Humanoid Robot Manufacturing Innovation Center releases the world’s first full-size humanoid robot open-source community OpenLoong, jointly building humanoid robot data sets with Fudan University, Jiaotong University, Tongji University, and leading enterprises, promoting industrial collaborative innovation and data open ecosystem formation.
(6) Suzhou: Activating Demand with Scenario Lists, Building Complementary Industrial Clusters through Endowment Division
Suzhou has a complete artificial intelligence industry system, with revenue reaching 236.2 billion yuan in 2024, a year-on-year growth of over 20%, gathering 667 core enterprises and over 2,100 upstream and downstream enterprises, forming a full-chain system from the foundational layer to the application layer. During the same period, the scale of the robotics industry reached 139.5 billion yuan, with over 600 enterprises on the chain, including over 80 core enterprises in embodied intelligent robots, covering core components, body manufacturing, and system integration, successfully achieving mass production of the world’s first open-source HarmonyOS humanoid robot and landing the first humanoid robot production line in Jiangsu. Suzhou’s development experience lies in using “AI + vertical industry” as a breakthrough, constructing an innovation pattern of “scenario-driven + regional collaboration.” First, leverage a solid manufacturing foundation to promote scenario-driven innovation through list management, releasing four batches of over 80 “city + AI” scenario lists, with 20 new key scenarios added in 2024, covering six major fields such as intelligent connected vehicles, smart healthcare, and intelligent manufacturing; simultaneously compiling the “Robot + Typical Scenario Application Directory,” directing enterprises’ technology iteration and product validation. Second, rely on regional endowment differences to layout the industrial chain, forming a complementary cluster system, coordinating ten major sectors: Suzhou Industrial Park focuses on cutting-edge directions such as AI large models, machine vision, and embodied intelligence, creating a core area for digital industries; Wuzhong District accelerates the extension of the “AI +” chain based on the advantages of the robotics and artificial intelligence trillion-level industrial cluster, approved as a provincial pilot for future humanoid robot industries. Suzhou achieves deep integration of AI and manufacturing through a model of “sector collaboration, scenario first,” constructing a nationally leading embodied intelligence innovation ecosystem.
4. Recommendations for Promoting the Development of the Artificial Intelligence and Robotics Industry through Integration and Innovation(1) Industry Side: Strengthen Key Links and Build Full-Chain Competitiveness
First, systematically focus on solidifying the data and computing power foundation, constructing a data ecosystem that adapts to the artificial intelligence and robotics industry. By formulating data collection standards, deploying intelligent sensors in key areas such as manufacturing and logistics, achieving automated collection of multimodal data; relying on platforms to develop integrated tools for “cleaning – labeling – desensitization,” establishing a dynamic quality assessment mechanism, forming a “general + specialized” industry corpus and high-quality data sets. Promote the construction of a multi-faceted data market of “platform + institution,” achieving mutual recognition and communication between industry data circulation platforms and trading institutions, and piloting trusted data spaces in key industries, creating industry-level/enterprise-level data training grounds to promote vertical scenario applications. Strengthen computing power supply and scheduling, constructing a “central + edge” collaborative architecture, upgrading artificial intelligence supercomputing centers, laying out regional edge computing centers and micro-computing nodes, establishing regional computing resource pools and unified scheduling platforms to support heterogeneous computing power integration and hybrid cloud elastic invocation. Cultivate large models in vertical fields, establish industry R&D centers and industry large model innovation centers, adopting a “enterprise posing questions + institutions tackling + scenario validation” model, developing a “general large model + industry knowledge graph” dual-engine architecture, promoting the “thousand enterprises thousand models” and “thousand models empowering hundreds of industries” projects, achieving breakthroughs in industrial manufacturing, port and shipping services, healthcare, and wellness, and accelerating the application of local innovative solutions through model validation zones and first purchase policies.
Second, accelerate breakthroughs in core products, strengthening the collaborative development of hardware, software, and complete systems. Focusing on key components such as high-end chips, power batteries, high-performance materials, connectors, and standard components, formulate a technology breakthrough list, promoting batch breakthroughs, supporting chip enterprises in R&D for edge computing, industrial control, and cloud training scenarios, optimizing energy efficiency and reliability indicators. Upgrade multimodal intelligent sensor technologies, develop specialized modules to achieve deep integration of hardware and control systems, enhancing adaptability in industrial manufacturing, logistics operations, etc., while promoting the localization of core components and the construction of shared trial production platforms, accelerating sample validation and iteration. Build an intelligent software ecosystem, developing real-time operating systems and industrial algorithm open-source platforms, creating a low-code manufacturing APP tool library, achieving hardware compatibility, industrial software adaptation, and vertical scenario optimization. Promote breakthroughs in complete systems, establishing a humanoid robot full-chain collaborative mechanism, supporting the priority procurement of automated production lines and local core components, promoting the soft-hard integration of intelligent connected vehicles, and developing special equipment such as explosion-proof and high-temperature resistant gear. Innovate AI consumer terminal products, covering smart homes, intelligent healthcare, audiovisual terminals, and wearable devices, integrating traditional Chinese medicine large models, immersive experiences, and flexible sensing technologies, achieving applications in home healthcare, health management, and extreme operational scenarios.
(2) Innovation Side: Build an Open Ecosystem to Stimulate Collaborative Innovation Vitality
First, strengthen the cultivation of integration innovation-type enterprises, constructing a multi-level enterprise echelon and collaborative development pattern. Promote the integration of artificial intelligence with high-end equipment manufacturing, supporting cross-border cooperation between complete machines and integrators, accelerating the upgrade iteration towards embodied intelligence. Focus on cultivating three types of enterprises: First, AI-native innovative enterprises, building technological barriers through self-developed algorithms, data closed loops, and open-source ecosystems; Second, scenario-deepening enterprises, supporting small and medium-sized enterprises in specialized development in segmented fields, cultivating “specialized, refined, and new small giants,” “single champions,” and “unicorns”; Third, ecosystem-building enterprises, promoting vertical resource integration, cultivating leading enterprises with ecological dominance and industry influence. Promote collaborative development of the industrial chain, accelerating R&D of core components and mass production of complete machines, constructing characteristic advantage industrial clusters, forming a closed loop of “complete machine traction + component matching.”
Second, strengthen technological open innovation, enhancing the overall technological capability and collaborative level of the artificial intelligence and robotics industry. Focus on breakthroughs in core common technologies such as perception decision-making, motion control, soft-hard collaboration, embodied corpus, and autonomous operating systems, implementing key area R&D plans, building digital twin training grounds, and promoting enterprises and research institutions to lead the formulation of international, national, and industry standards. Construct a comprehensive open-source ecosystem, supporting enterprises, universities, research institutions, and industry associations to jointly build open-source communities, ecological centers, and public service platforms, forming a comprehensive open-source system of models, data, algorithms, operating systems, and toolchains, providing technical sharing, ecological promotion, computing power scheduling, big data training libraries, and large model evaluation and verification services. Break through the trial conversion chain of production, learning, and research, promoting various innovative entities to tackle challenges together, forming industrial innovation alliances and laboratories, trial production bases, accelerating the collaborative R&D of foundational bases and core products, and focusing on supporting the construction of embodied intelligence trial production platforms, providing integrated trial services before large-scale mass production, achieving efficient connections from technology R&D to industrial application.
Third, strengthen platform support capabilities, enhancing the overall efficiency of R&D, validation, and application in the artificial intelligence and robotics industry. Build a full-domain heterogeneous humanoid robot training base, integrating multi-brand and multi-form robots for collaborative training, driving the evolution of embodied intelligence through large-scale data collection, constructing cross-vendor data sharing mechanisms and general large model foundations, supporting enterprises to call open-source algorithm libraries and training data sets, reducing R&D costs, and building specialized training grounds to achieve rapid iteration of data feedback in industrial scenarios. Construct a rapid manufacturing development and testing evaluation platform, laying out an integrated production line of “design – trial production – mass production,” providing rapid prototyping and process optimization for core components such as high-precision joints and bionic dexterous hands, while establishing functional safety, environmental adaptability, and human-machine interaction testing and evaluation systems, promoting industry standard certification. Create scenario-based training grounds and industrial ecosystem collaboration platforms, relying on manufacturing advantages to build digital twin training grounds, jointly constructing open-source communities with universities and key laboratories, opening AR/VR simulation toolchains and training data sets, supporting scenario insurance funds and leasing subsidies, forming a positive cycle of “training data – technology optimization – scenario validation – commercial promotion,” lowering the access threshold for small and medium-sized enterprises, accelerating technology landing and industrial application.
(3) Operation Side: Improve Industrial Carriers and Enhance Industry Influence
First, create specialized industrial parks and industry landmarks. Relying on regions with a foundation for complete machine manufacturing, build specialized industrial parks for humanoid robots, embodied intelligence, core components, and intelligent systems, promoting efficient connections in the industrial chain through collaboration between complete machines and components. Promote the construction of a secondary development industrial community for robots, empowering software and hardware integration, application scenario expansion, and sharing of innovative results. Encourage planning and construction of artificial intelligence characteristic industrial spaces and buildings, focusing on integrated development of “key technologies + core products + application scenarios,” laying out functions such as R&D design, data training, computing power sharing, and trial application, building comprehensive robot industry carriers that integrate R&D, production, and validation. Relying on core scientific innovation zones, create landmarks for the artificial intelligence + robotics industry, establishing pilot areas for integrated innovative applications.
Second, strengthen scenario-based standards and international rule output. Relying on special actions for standard internationalization, promote local standards to upgrade to national/international standards, and establish standard promotion funds, releasing multilingual technical white papers at international exhibitions, exporting intelligent solutions to countries along the Belt and Road. Provide special subsidies to enterprises leading international standards, encouraging enterprises to participate in the work of international robotics technical committees, enhancing their voice in the formulation of global robotics rules. Construct a cross-border mutual recognition pool for standards, prioritizing international cooperation projects, shortening mutual recognition cycles, and providing international certification services relying on open laboratories, strengthening institutional open tools support.
Third, create a batch of high-end communication platforms for the industry. Organize enterprises to actively participate in the World Artificial Intelligence Conference, World Robotics Conference, Global Developer Pioneer Conference, National Robotics Challenge, and other conferences and competitions, building global cooperation platforms and collaborative networks. Explore creating influential national brand activities focused on artificial intelligence and robotics, supporting the holding of artificial intelligence + innovation competitions, technology open days, entrepreneurship salons, and other activities, promoting strengthened connections and cooperation among upstream and downstream of the industrial chain, large, medium, and small enterprises.
(4) Element Side: Gather High-End Resources and Provide Comprehensive Support
First, attract and cultivate a group of artificial intelligence service providers. Jointly compile and release a development map for artificial intelligence and robotics service providers with relevant departments, focusing on attracting and cultivating a group of model service providers, data service providers, and industry AI application solution enterprises, guiding existing service providers towards specialization and differentiation. Establish a gradient cultivation system for service providers, jointly forming an “AI service provider application alliance” with leading enterprises, supporting leading enterprises to jointly build AI service empowerment centers.
Second, attract and cultivate a team of industry composite talents. Support enterprises in attracting and cultivating leading talents and young top talents with technological innovation capabilities and the ability to solve complex engineering problems, promoting local universities to set up interdisciplinary majors in embodied intelligent robotics, cultivating composite talents with capabilities in artificial intelligence, mechanical engineering, and automation. Encourage joint training classes, scenario practices, and specialized training between schools and enterprises, constructing a talent echelon of “leading talents tackling challenges + practical craftsmen supporting.” Implement preferential policies for high-level talents, providing long-term career development and funding support, while innovating talent evaluation standards, incorporating salary levels, open-source contributions, and training optimization experiences into key indicators, enhancing the attractiveness of high-end talents and filling the talent gap in the fields of humanoid robotics and artificial intelligence.
Third, form a batch of financial innovation platform support. Fully leverage the guiding role of state-owned capital, establish industrial investment funds, and support seed and startup projects in AI large models, core components of humanoid robots, etc., through market-oriented mechanisms such as jointly establishing special sub-funds and collaborative investments, establishing a nurturing mechanism for patient capital covering “technology R&D – trial verification – scenario landing” and a gradient exit channel, creating a characteristic fund ecosystem of “early investment, small investment, long-term investment, and new investment.” Smooth credit financing channels, linking banks, insurance, and other institutions to innovate and establish special products such as “computing power loans” and “artificial intelligence loans (insurance),” and establish risk compensation pools and insurance funds for first sets of equipment, providing risk protection for intellectual property pledges.
