《【中英文对照版】人形机器人创新发展指导意见.docx》由会员分享，可在线阅读，更多相关《【中英文对照版】人形机器人创新发展指导意见.docx（27页珍藏版）》请在课桌文档上搜索。
1、（工信部科（2023 ） 193 号）各省、自治区、直辖市及计 划单列市、新疆生产建设兵 团工业和信息化主管部门， 有关行业协会、企事业单 位：现将人形机器人创新发展 指导意见印发给你们，请 结合实际，认真贯彻落实。人形机器人创新发展指导意见GuidingOpinionsontheInnovativeDevelopmentofHumanoidRobots制定机关：工业和信息化部发文字号：工信部科2023193号公布日期：2023.10.20施行日期：2023.10.20效力位阶：部门规范性文件法规类别：机械工业管理IssuingAuthority：MinistryofIndustry&Infb
2、nnationTechnologyDocumentNumber：No.1932023oftheMiniStryOflndUSIryandInfonnationTechnologyDateIssued：10-20-2023EffectiveDale:10-20-20231.evelofAuthority：DepartmentalRegulatoryDocumentsAreaofLaw：MachineryIndustryAdministrationNoticebytheMinistryofIndustryandInformation工业和信息化部关于印发TechnologyofIssuingthe
3、GuidingOpinionson人形机器人创新发展指导theInnovativeDevelopmentofHumanoidRobots意见的通知(No.1932023oftheMinistryofIndustryandInformationTechnology)Theindustryandinformationtechnologyauthoritiesofallprovinces,autonomousregions,municipalitiesdirectlyundertheCentralGovernment,citiesunderseparatestateplanning,andXinji
4、angProductionandConstructionCorps;andrelevantindustryassociations,enterprises,andpublicorganizations:TheGuidingOpinionsontheInnovativeDevelopmentofHumanoidRobotsareherebyissuedforyourimplementationinlightofactualconditions.MinistryofIndustryandInformationTechnology工业和信息化部October20,20232023年10月20日Ann
5、ex:GuidingOpinionsontheInnovativeDevelopmentofHumanoidRobotsGuidingOpinionsontheInnovativeDevelopmentofHumanoidRobotsHumanoidrobots,anintegrationofadvancedtechnologiessuchasartificialintelligence,high-endmanufacturing,andnewmaterials,areexpectedtobecomedisruptiveproductsfollowingcomputers,smartphone
6、s,andnewenergyvehicles.Theywillprofoundlychangethewayofproductionandlifeandreshapetheglobalindustrialdevelopmentlandscapes.Ashumanoidrobottechnologyundergoesacceleratingevolution,itbecomesanewcontestedfrontieroftechnology,ushersinfutureindustries,andfuelseconomicdevelopment.Ithasgreatpotentialfordev
7、elopmentandextensiveapplication.TheseGuidingOpinionsareformulatedforthepurposesofpromotingthehigh-qualitydevelopmentofthehumanoidrobotindustry,cultivatingnewproductiveforces,enablinghigh-levelnewindustrialization,andstronglysupportingtheconstructionofamodernindustrialsystem.I.Generalthinking1. Guidi
8、ngideology附件：人形机器人创新发展 指导意见人形机器人创新发展指导意 见人形机器人集成人工智能、 高端制造、新材料等先进技 术，有望成为继计算机、智 能手机、新能源汽车后的颠 覆性产品，将深刻变革人类 生产生活方式，重塑全球产 业发展格局。当前，人形机 器人技术加速演进，已成为 科技竞争的新高地、未来产 业的新赛道、经济发展的新 引擎，发展潜力大、应用前 景广。为推动人形机器人产 业高质量发展，培育形成新 质生产力，高水平赋能新型 工业化，有力支撑现代化产 业体系建设，制定本指导意 见。一、总体思路（一）指导思想以习近平新时代中国特色社 会主义思想为指导，深入贯 彻落实党的二十大精神
9、，完GuidedbyXiJinpingThoughtonSocialismwithChineseCharacteristicsforaNewEra,wewillthoroughlyimplementtheguidingprinciplesofthe20thCPCNationalCongress,fullyandfaithfullyapplythenewdevelopmentphilosophyonallfronts,createanewdevelopmentdynamic,andoverallplandevelopmentandsecurity.Guidedbybreakthroughsina
10、rtificialintelligencetechnologies,suchaslargelanguagemodels(LLMs)5andbuildinguponmatureroboticstechnology,wewilladheretothepathofapplication-drivendevelopment,impetusfromcompletelybuiltunits,coordinationbetweenhardwareandsoftware,andecosystemcultivation.Bytechnologygrading,generationalevolutionofpro
11、ducts,andtaskphasing,weW川leveragetheadvantagesofacomprehensivemanufacturingsector,diverseapplicationscenarios,avastmarket,andthenewnationalsystem.Bydoingso,wewillpromotetheinnovativedevelopmentofChina,shumanoidrobotindustry,supportingtheinitiativestoenhancemanufacturingandcyberspacecapabilitiesandbu
12、ildadigitalChina.2. Developmentgoals整、准确、全面贯彻新发展 理念，加快构建新发展格 局，统筹发展和安全，以大 模型等人工智能技术突破为 引领，在机器人已有成熟技 术基础上，坚持应用牵引、 整机带动、软硬协同、生态 培育的路径，采取技术分 级、产品分代、任务分期的 方式，发挥制造业门类齐 全、应用场景丰富、市场规 模庞大以及新型举国体制优 势，加快推动我国人形机器 人产业创新发展，为建设制 造强国、网络强国和数字中 国提供支撑。（二）发展目标到2025年，人形机器人创新 体系初步建立，“大脑、小 脑、肢体”等一批关键技术 取得突破，确保核心部组件 安全有效供
13、给。整机产品达 到国际先进水平，并实现批 量生产，在特种、制造、民 生服务等场景得到示范应 用，探索形成有效的治理机 制和手段。培育2-3家有全 球影响力的生态型企业和一 批专精特新中小企业，打造 2-3个产业发展集聚区，孕 育开拓一批新业务、新模 式、新业态。For2025,thegoalsincludetheinitialestablishmentofaninnovationsystemforhumanoidrobots,achievingbreakthroughsinkeytechnologiessuchasubrain,cerebellum,andIimbstoensurethesec
14、ureandeffectivesupplyofessentialcomponents.Thegoalsfurtherencompasselevatingcompletelybuiltunitstointernationalstandards,initiatingmassproduction,anddemonstratingtheirapplicationsinspecialized,manufacturing,publicservice,andotherscenarios.Concurrently,effortswillbedirectedtowardsexploringeffectivego
15、vernancemechanismsandmethods.Additionally,theaimistocultivatetwotothreeecosystementerpriseswithglobalinfluenceandfostersmallandmedium-sizedspecializedenterprisesofferingnew,unique,andcompetitiveproducts.Furthermore,theaspirationincludesthecreationoftwotothreeindustrialdevelopmentclustersthatsupporta
16、nddevelopnewbusiness,models,andforms.Fortheyear2027,thegoalsinvolveasubstantialenhancementofthecapabilitytoinnovatehumanoidrobottechnology.Thisincludestheestablishmentofasecureandreliableindustrialchainandsupplychain,thecreationofaninternationallycompetitiveindustrialecosystem,andtheelevationofcompr
17、ehensivestrengthtoaglobalstandard.Additionally,thegoalsencompasslarge-scaleindustrialdevelopment,theproliferationofdiverseapplicationscenarios,andtheintegrationofrelatedproductsintotherealeconomy,therebyservingasasignificantnewdriverforeconomicgrowth.II.Breakthroughsinkeytechnologies1.Buildingthe“br
18、ain”and,cerebellumofhumanoidrobots到2027年，人形机器人技术 创新能力显著提升，形成安 全可靠的产业链供应链体 系，构建具有国际竞争力的 产业生态，综合实力达到世 界先进水平。产业加速实现 规模化发展，应用场景更加 丰富，相关产品深度融入实 体经济，成为重要的经济增 长新引擎。二、突破关键技术（一）打造人形机器人“大 脑”和“小脑”开发基于人工智能大模型的 人形机器人“大脑”，增强 环境感知、行为控制、人机 交互能力，推动云端和边缘 端智能协同部署。建设大模 型训练数据库，创新数据自 动化标注、清洗、使用等方 法，扩充高质量的多模态数 据。科学布局人形机器人
19、算 力，加速大模型训练迭代和 产品应用。开发控制人形机 器人运动的“小脑”，搭建 运动控制算法库，建立网络 控制系统架构。面向特定应 用场景，构建仿真系统和训TheLLM-basedbrainofhumanoidrobotsshallbedevelopedtoenhancecontextawareness,behavioralcontrol,andhuman-computerinteractioncapabilities.Itsintelligentcollaborativedeploymentoncloudandattheedgeshallbepromoted.AnLLMtrainingda
20、tabaseshallbebuilt,incorporatinginnovativemethodsforautomatedannotation,cleaning,anduseofdata,toexpandhigh-qualitymulti-modaldata.Computingpowerforhumanoidrobotsshallbescientificallylaidout,expeditingthetraininganditerationofLLMsandtheapplicationofproducts.Acerebellumthatcontrolsthemotionofhumanoidr
21、obotsshallbedevelopedbybuildingarepositoryformotioncontrolalgorithmsandanetworkedcontrolsystemarchitecture.For练环境，加快技术迭代速 度，降低创新成本。specificapplicationscenarios,simulationsystemsandtrainingenvironmentsshallbecreatedtospeeduptechnologyiterationandreduceinnovationcosts.2.Breakthroughsinkey,limbntechnol
22、ogies（二）突破“肢体”关键技术用好现有机器人技术基础， 系统部署“机器肢”关键技 术群，创新人体运动力学基 础理论，打造仿人机械臂、 灵巧手和腿足，突破轻量化 与刚柔耦合设计、全身协调 运动控制、手臂动态抓取灵 巧作业等技术。攻关“机器 体”关键技术群，突破轻量 化骨骼、高强度本体结构、 高精度传感等技术，研发高 集成、长续航的人形机器人 动力单元与能源管理技术。（三）健全技术创新体系构建完善人形机器人制造业 技术创新体系，凝练关键技 术、物料、企业、制造装 备、质量、标准、关键软件 等清单，精准推进“补短锻 长”。支持龙头企业牵头联 合产学研用组成创新联合 体，加强关键技术和产品攻 关
23、，凝聚各方力量加快创新 进程。加快人形机器人与元Buildingonexistingrobotictechnology,keytechnologyforroboticlimbsshallbesystematicallydeployed,andthefoundationalbiomechanicaltheoryonhumanmovementsshallbeinnovated,soastocreatemechanicalarmsanddexteroushands,legs,andfeetofhumanoidsandmakebreakthroughsinlightweightandrigid-fle
24、xcouplingdesign,coordinatedwhole-bodymotioncontrol,dynamicgraspingandpreciseoperationsbyarmsandhands,andothertechnologies.Keytechnologiesforrobotbodiesshallberesearchedtomakebreakthroughsintechnologiessuchaslightweightbones,high-strengthmechanicalparts,andhigh-precisionsensinganddevelophighlyintegra
25、ted,long-lastingpowerunitsofhumanoidrobots,andenergymanagementtechnologies.3.ImprovingthetechnologicalinnovationsystemAtechnologicalinnovationsystemforhumanoidrobotmanufacturingshallbeestablishedandimproved.Alistofkeytechnologies,materials,enterprises,manufacturingequipment,quality,standards,andkeys
26、oftware,amongothers,shallbecreatedtoaccuratelyaddressweaknessesandimprovestrengths.Leadingenterprisesshallbesupportedintakingtheleadinforminganinnovationconsortiumwithotherenterprises,educationalinstitutions,researchinstitutes,andusers,soastoresearchkeytechnologiesand宇宙、脑机接口等前沿技术 融合，探索跨学科、跨领域 的创新模式。
27、专栏1关键技术攻关?productsandcombinetheeffortsofallstakeholderstoacceleratetheinnovationprocess.Humanoidrobotsshallbeintegratedwithcutting-edgetechnologiessuchasthemetaverseandbrain-computerinterfacetoexploreinterdisciplinaryandcross-fieldinnovationmodels.机器人“大脑”关键技术 群。围绕动态开放环境下 人形机器人感知与控制， 突破感知-决策-控制一体 化的端
28、到端通用大模型、 大规模数据集管理、云边 端一体计算架构、多模态 感知与环境建模等技术， 提高人形机器人的人-机- 环境共融交互能力，支撑 全场景落地应用。机器人“小脑”关键技术 群。面向人形机器人复杂 地形通过、全身协同精细 作业等任务需求，开展高 保真系统建模与仿真、多 体动力学建模与在线行为 控制、典型仿生运动行为 表征、全身协同运动自主 学习等关键技术研究，提 升人形机器人非结构化环 境下全身协调鲁棒移动、 灵巧操作及人机交互能 力。机器肢关键技术群。面向 人形机器人高动态、高爆 发和高精度等运动性能需 求.研究人体力学持征及Box1KeytechnologyresearchKeyte
29、chnologiesforrobot,brain:Withafocusonthesenseandcontrolofhumanoidrobotsinadynamicopenenvironment,breakthroughsshallbemadeintechnologies,suchasend-to-endgeneral-purposeLLMsthatintegratesensing,decision-making,andcontrol,large-scaledatasetmanagement,cloud-edge-deviceintegratedcomputingarchitecture,mul
30、ti-modalsensing,andenvironmentmodeling,amongothers.Thesebreakthroughsaimtoimprovethehumanoidrobots*capabilityofintegratingandinteractingwithhumanandtheenvironmentandsupporttheirapplicationsinallscenarios.Keytechnologiesforrobotcerebellum:Tomeettherequirementsthathumanoidrobotspassthroughcomplexterra
31、inandperformprecisewhole-bodycoordinatedoperationsandothertasks,keytechnologiessuchashigh-fidelitysystemmodelingandsimulation,multibodysystemmodelingandonlinebehavioralcontrol,characterizationoftypicalbiologicallyinspiredbehaviors,andautonomouslearningofwhole-bodycoordinatedmotion,shallberesearchedt
32、oimprovehumanoidrobot,sfullbodycoordinatedandrobustmovement,dexterousoperation,andhuman-computerinteractioncapabilitiesinunstructuredenvironments.Keytechnologiesforroboticlimbs:Tomeettheneedsofhumanoidrobotsfordynamic,bursting,precision,andothermotionperformance,basictheoriesonhumanmechanicalcharact
33、eristicsandmotionmechanismsandthedynamicmodelsandcontrolofhumanoidrobotsshallberesearched,andbreakthroughsshallbemadeinrigid-flexiblecouplingsforbiologicallyinspiredmechanicalpowertransmission,highlycompactrobotlimbstructures,dexteroushanddesign,andotherkeytechnologies,soastolayasolidhardwarefoundat
34、ionforflexiblemovementofhumanoidrobots.Keytechnologiesforrobotbodies:Tomeetthehigh-strengthandcompactstructuralrequirementsforhumanoidrobotsmechanicalparts,researchshallbeconductedonartificialintelligence-driventopologyoptimizationofskeletalstructure,high-strengthandlightweightnewmaterials,additivem
35、anufacturingofcomplexbodystructure,energy-structuresensingintegrateddesign,andprotectioninharshenvironmenttocreateastructureforhumanoidrobotsmechanicalpartscharacterizedbyhighsafety,highreliability,andhighenvironmentaladaptability.运动机理、人形机器人动 力学模型及控制等基础理 论，突破刚柔耦合仿生传 动机构、高紧凑机器人四 肢结构与灵巧手设计等关 键技术，为人形机器
36、人灵 活运动夯实硬件基础。机器体关键技术群。面向 人形机器人本体高强度和 高紧凑结构需求，研究人 工智能驱动的骨架结构拓 扑优化、高强度轻量化新 材料、复杂身体结构增材 制造、能源-结构-感知一 体化设计以及恶劣环境防 护等关键技术，打造具有 高安全、高可靠、高环境 适应性的人形机器人本体 结构。？III. Cultivatingkeyproducts三、培育重点产品1. Creatingcompletehumanoidrobotproducts（一）打造整机产品打造基础版整机，构筑人形 机器人通用整机平台，支持 后续个性化功能开发。面向 不同应用场景需求，开发低 成本交互型、高精度型以及 极
37、端环境下高可靠型等人形 机器人整机产品Q针对不同 动力需求，推出电驱动、液 压驱动或混合驱动人形机器Abasicversionofcompletelybuiltunitsshallbecreated,formingageneral-purposeplatformforhumanoidrobotstosupportthesubsequentdevelopmentoftailoredfunctions.Tomeettheneedsofvariousapplicationscenarios,therewillbeafocusondevelopinglow-cost,interactive,high-
38、precisioncompletelybuilthumanoidrobotproductsthatdemonstratehighreliabilityinextremeenvironments.Tomeetdifferentpowerrequirements,electric-driven,hydraulic-driven,or人。强化人形机器人整机的 批量化生产制造能力，持续 提升整机产品的质量和可靠 性。（二）夯实基础部组件hybrid-drivenhumanoidrobotsshallbelaunched.Thecapabilitytomass-producehumanoidrobots
39、shallbeenhanced,andthequalityandreliabilityofcompletebuiltunitsshallcontinuetobeimproved.2. Consolidatingbasiccomponents聚焦人形机器人专用传感 器，突破视、听、力、嗅等 高精度传感关键技术，提升 环境综合感知能力。发展高 功率密度执行器，满足本体 高爆发移动和高精度作业需 求。开发面向人形机器人的 专用芯片，提升运动控制、 认知决策等计算效能。结合 新能源产品发展趋势，研发 适用于人形机器人特点的高 能效专用动力组件。Withafocusonsensorsforhumanoi
40、drobots,breakthroughsshallbemadeinkeyhigh-precisionsensingtechnologiessuchasvision,hearing,force,andsmelltoimprovecomprehensivecontextawarenesscapabilities.High-power-densityactuatorsshallbedevelopedtomeettheneedsforintenseburstsofmovementandhigh-precisionoperationsofmechanicalparts.Chipsforhumanoid
41、robotsshallbedevelopedtoimprovecomputingperformanceformotioncontrolandcognitivedecision-making,amongothers.Basedonthedevelopmenttrendofnewenergyproducts,high-efficiencypowercomponentsinlinewiththecharacteristicsofhumanoidrobotsshallbedeveloped.（三）推动软件创新3. Promotingsoftwareinnovation构建人形机器人高实时、高 可靠、高
42、智能的专用操作系 统，推动与通用大模型等技 术深度融合，提供安全、稳 定、易用的系统平台。开发 面向各类场景的应用软件， 建设完善人形机器人应用开 发平台和工具包，构建丰富 的软件赋能应用生态。探索 “机器人即服务”的新模 式，加速人形机器人低成 本、灵活部署应用。Tomeettherequirements,adedicatedoperatingsystemforhumanoidrobotswithhighreal-time,highreliability,andhighintelligenceshallbedeveloped.ThissystemW川bedeeplyintegratedwit
43、hgeneralLLMandothertechnologies,aimingtoprovideasafe,stable,anduser-friendlyenvironment.Developmentofapplicationsoftwareforvariousscenariosshallbedeveloped,alongsidethecreationandimprovementofhumanoidrobotapplicationdevelopmentplatformsandtoolkits,withtheaimofbuildingarichsoftware-enabledapplication
44、ecosystem.Anewmodelofrobotsasaservicenshallbeexploredforthelow-cost,flexibledeploymentandapplicationofhumanoidrobots.Box2:ResearchonkeyproductsandpartsBasicversionofcompletelybuiltunits:Inpursuitofbasicformsandfunctionalitiessuchashumanoidappearance,bipedality,anddexterousoperationsofarmsandhands,ab
45、asichardwareandsoftwarearchitectureforhumanoidrobotsshallbeestablished,andapublicversionofgeneral-purposeplatformsshallbecreatedtosupportstructuralmodifications,algorithmoptimizations,andspecificcapabilityenhancementstomeetdiversescenariorequirements.专栏2重点产品和部组件 攻关基础版整机。面向类人外 观、双腿行走和双臂双手 灵巧操作的基本形态功
46、能，建立人形机器人基础 软硬件架构，打造“公 版”通用平台，支持不同 场景需求下的结构改造、 算法优化以及特定能力强 化。功能型整机。开发低成本 交互型人形机器人，强化 人类生活环境适应能力、 多模态人机交互能力。开 发高精度型人形机器人， 强化双臂双手精细操作、 工件鲁棒识别、轨迹智能 规划等上肢作业能力。开 发高可靠型人形机器人， 强化恶劣环境生存、复杂 地形适应、外力冲击防护 等能力。传感器。面向复杂环境感 知需求，开发集成高精度 仿生眼与类脑处理算法的 视觉传感器，推出宽频 响、高灵敏的仿生听觉传 感器，开发高分辨率和具 有多点接触检测能力的仿 人电子皮肤，推出高灵敏 检测多种气体的仿
47、生嗅觉 传感器，形成人形机器人 专用传感器产品谱系CFunctionalcompletelybuiltunits:Low-cost,interactivehumanoidrobotsshallbedevelopedtoenhancetheiradaptabilitytothehumanlivingenvironmentandcapabilityofmulti-modalhuman-computerinteraction.High-precisionhumanoidrobotsshallbedevelopedtoenhanceupper-limboperationcapabilitiessuch
48、aspreciseoperationofarmsandhands,robustidentificationofworkpieces,andintelligenttrajectoryplanning.High-reliabilityhumanoidrobotsshallbedevelopedtostrengthentheircapabilitytosurviveinharshenvironments,adapttocomplexterrain,andwithstandtheimpactsofexternalforces.Sensors:Inordertomeetthedemandsofcompl
49、excontextawareness,visionsensorsshallbedeveloped,integratinghigh-precisionbiologicallyinspiredeyeswithneuromorphicprocessingalgorithms.Additionally,highlysensitivebiologicallyinspiredauditorysensorswithawide-rangefrequencyresponseshallbelaunched.Thedevelopmentofhigh-resolutionelectronicskinwithmulti-pointtouchdetectioncapabilitiesshallbepursued,alongwiththelaunchofbiol
