Definition
Weunderstandtheso-calledintelligentrobotinabroadsense.Themostimpressivethingitgivespeopleisaunique"livingcreature"thatperformsself-control.Infact,themainorganofthisself-control"livingthing"isnotassubtleandcomplicatedasarealperson.
Intelligentrobotshavevariousinternalandexternalinformationsensors,suchassight,hearing,touch,andsmell.Inadditiontosusceptors,italsohaseffectorsasameanstoactonthesurroundingenvironment.Thisisthemuscles,orself-regulatingmotors,whichmakethehands,feet,longnose,tentacles,etc.move.Itcanalsobeseenthatintelligentrobotsmusthaveatleastthreeelements:sensoryelements,reactionelementsandthinkingelements.
Intelligentrobot(12photos)
Wecallthiskindofrobotaself-controlrobotinordertodistinguishitfromtherobotsmentionedabove.Itistheresultofcybernetics,whichadvocatesthefactthatlifeandnon-lifepurposefulbehaviorareconsistentinmanyrespects.Asanintelligentrobotmakersaid,robotsareafunctionaldescriptionofasystem.Thiskindofsystemcanonlybeobtainedfromtheresultsofthegrowthoflivingcellsinthepast.Theyhavebecomethingsthatwecanmakeourselves.Theintelligentrobotcanunderstandhumanlanguage,usehumanlanguagetotalktotheoperator,andinitsown"consciousness"formsasingleexternalenvironmentthatallowsitto"live"-adetailedmodeloftheactualsituation.Itcananalyzethesituation,adjustitsactionstomeetalltherequirementsputforwardbytheoperator,drawupthedesiredactions,andcompletetheseactionsundertheconditionsofinsufficientinformationandrapidchangesintheenvironment.Ofcourse,itisimpossibletomakeitexactlythesameasourhumanthinking.However,therearestillpeopletryingtobuildsomekindof"microworld"thatcomputerscanunderstand.
Functionalclassification
Canbedividedintogeneralrobotsandintelligentrobots.
Generalrobotsrefertorobotsthatdonothaveintelligence,butonlyhavegeneralprogrammingcapabilitiesandoperatingfunctions.
Thereisnounifieddefinitionofintelligentrobotsintheworld.Mostexpertsbelievethatintelligentrobotsmusthaveatleastthefollowingthreeelements:oneisthesensoryelement,whichisusedtorecognizethestateofthesurroundingenvironment;thesecondisthemovementelement,whichmakesreactiveactionstotheoutsideworld;thethirdisthethinkingelement,whichisobtainedfromthesensoryelement.Information,thinkaboutwhatkindofactiontotake.Sensoryelementsincludenon-contactsensorsthatcanperceivevision,proximity,distance,etc.,andcontactsensorsthatcanperceivepower,pressure,andtouch.Theseelementsareessentiallyequivalenttohumaneyes,nose,earsandotherfivesenses.Theirfunctionscanuseelectromechanicalelementssuchascameras,imagesensors,ultrasonictransducers,lasers,conductiverubber,piezoelectriccomponents,pneumaticcomponents,andtravelswitches.Devicetoachieve.Formotionelements,intelligentrobotsneedatracklessmobilemechanismtoadapttodifferentgeographicenvironmentssuchasflatground,steps,walls,stairs,andramps.Theirfunctionscanbeaccomplishedwiththehelpofmovingmechanismssuchaswheels,crawlers,feet,suctioncups,andaircushions.Real-timecontrolofthemovingmechanismisrequiredduringexercise.Thiscontrolincludesnotonlypositioncontrol,butalsoforcecontrol,positionandforcemixedcontrol,andexpansionratecontrol.Thethinkingelementoftheintelligentrobotisthekeyofthethreeelements,anditisalsotheessentialelementthatpeoplewanttogivetherobot.Thinkingelementsincludeintellectualactivitiessuchasjudgment,logicalanalysis,andunderstanding.Theseintellectualactivitiesareessentiallyaninformationprocessingprocess,andthecomputeristhemainmeanstocompletethisprocess.Intelligentrobotscanbedividedintothreetypesaccordingtotheirintelligencelevel:
Sensingtype
Alsoknownasexternallycontrolledrobots.Thereisnointelligentunitonthebodyoftherobot,onlytheactuatorandthesensingmechanism.Ithastheabilitytousesensorinformation(includingvision,hearing,touch,proximity,forceandinfrared,ultrasoundandlaser,etc.)toprocesssensorinformation,realizecontrolandTheabilitytooperate.Itiscontrolledbyanexternalcomputerandhasanintelligentprocessingunitontheexternalcomputertoprocessvariousinformationcollectedbythecontrolledrobotandvariousposturesandtrajectoriesoftherobotitself,andthenissuecontrolinstructionstodirecttherobot'sactions.TherobotsusedintheRobotWorldCupsmallgroupmatchesbelongtothistype.
Interactivetype
Therobotconductsman-machinedialoguewiththeoperatororprogrammerthroughthecomputersystemtorealizethecontrolandoperationoftherobot.Althoughithassomeprocessinganddecision-makingfunctions,itcanindependentlyrealizesomefunctionssuchastrajectoryplanningandsimpleobstacleavoidance,butitisalsosubjecttoexternalcontrol.
Autonomoustype
Afterdesignandproduction,therobotdoesnotrequirehumaninterventionandcanautomaticallycompletevariousanthropomorphictasksinvariousenvironments.Theautonomousrobothasmodulesofperception,processing,decision-making,execution,etc.,andcanindependentlymoveanddealwithproblemslikeanautonomousperson.Therobotsusedinthemedium-sizedcompetitionsoftheRobotWorldCupbelongtothistype.Themostimportantfeatureofafullyautonomousmobilerobotisitsautonomyandadaptability.Autonomymeansthatitcanperformcertaintaskscompletelyautonomouslyinacertainenvironmentwithoutrelyingonanyexternalcontrol.Adaptabilitymeansthatitcanrecognizeandmeasuresurroundingobjectsinrealtime,adjustitsownparameters,adjustactionstrategies,andhandleemergenciesaccordingtochangesintheenvironment.Interactivityisalsoanimportantfeatureofautonomousrobots.Robotscancommunicateinformationwithpeople,withtheexternalenvironment,andwithotherrobots.Sincefullyautonomousmobilerobotsinvolveresearchinmanyaspectssuchasdrivecontrol,sensordatafusion,imageprocessing,patternrecognition,neuralnetworks,etc.,theycancomprehensivelyreflectacountry'slevelofmanufacturingandartificialintelligence.Therefore,manycountriesattachgreatimportancetotheresearchoffullyautonomousmobilerobots.
Theresearchofintelligentrobotsbeganintheearly1960s.Afterdecadesofdevelopment,intelligentrobotsbasedonsensorycontrol(alsoknownassecond-generationrobots)havereachedthepracticalapplicationstage.Intelligentrobotsbasedonknowledgecontrol(Alsoknownasautonomousrobotsornext-generationrobots)havealsomadegreatprogress,andavarietyofprototypeshavebeendeveloped.
Classificationofintelligencelevel
Industrialrobot
Itcanonlyworkrigidlyinaccordancewiththeproceduresprescribedbypeople.Nomatterwhatchangesinexternalconditions,itcannotbedonebyitself.Theprocedureistomakecorrespondingadjustmentstotheworkdone.Ifyouwanttochangetheworkdonebytherobot,humansmustmakecorrespondingchangestotheprogram,soitisnotintelligent.
ElementaryIntelligence
Itisdifferentfromindustrialrobots.Ithastheabilitytofeel,recognize,reasonandjudgelikehumans.Theprogramcanbemodifiedwithinacertainrangeaccordingtochangesinexternalconditions,thatis,itcanadapttochangesinexternalconditionsandadjustitselfaccordingly.However,theprincipleofthemodificationprocedureisprescribedbythepersoninadvance.Thiskindofprimaryintelligentrobotalreadypossessesacertaindegreeofintelligence.Althoughithasnoautomaticplanningability,thiskindofprimaryintelligentrobothasalsobeguntomatureandreachapracticallevel.
Intelligentagriculture
Theshark-typeintelligentagriculturalrobotadoptsaerodynamicsandformsashark-shapedappearancestructureaccordingtothecharacteristicsofaerodynamiclayout.Itusesacrawler-typechassismadeofindustrial-gradepolymermaterials.Theangledesignofthedepartureanglecanensurethattherobotisunimpededinvariouscomplexterrainorchards,andprotectthefarmlandfromdamage;theuniquemechanicaldesigncombinedwiththestreamlinedstructurecanmaximizetheuseofequipmentspace,andthemaximumloadcapacityisupto600kg;doubleThelayoutoftheengineensurestherobot'sgoodworkingability.Theuseoffly-by-wirecontroltechnologycombinedwiththeself-developedhydraulicsystemenablestherobottobreakthroughtheproblemofshortenduranceandhavesuperlongendurance;itadopts300MVHFwirelessremotecontroland5.8GimageThetransmissiontechnologycanimplementtherunningdataandimagesoftheinspectionproduct,andcanplanthepathattheterminal,whichcantrulyrealizeautomaticcontrol,andcanquicklyrealizefunctionexpansionandproductinnovation;theintelligentspraysystemdirectionalcapturesthecanopyoffruittrees
FamilyIntelligentEscort
Theescortrobotisusedintheenvironmentofnursinghomesorcommunityservicestations,andhasfunctionssuchasphysiologicalsignaldetection,voiceinteraction,telemedicine,intelligentchat,andautonomousobstacleavoidanceroaming.
Therobotrealizesthefunctionofautonomousnavigationandobstacleavoidanceinthenursinghomeenvironment,andcaninteractthroughvoiceandtouchscreen.Withrelateddetectionequipment,therobothasthefunctionsofdetectingandmonitoringphysiologicalsignalssuchasbloodpressure,heartbeat,bloodoxygen,etc.Itcanwirelesslyconnecttothecommunitynetworkandtransmitittothecommunitymedicalcenter.Incaseofemergency,itcanpromptlyalarmornotifyrelatives.Therobothasanintelligentchatfunction,whichcanassisttheelderlyinpsychologicalrehabilitation.Theescortrobotprovidessolutionstomajorsocialproblemscausedbytheagingofthepopulation.
AdvancedIntelligence
Sameaselementaryintelligentrobots,advancedintelligentrobotshavetheabilitytosense,recognize,reasonandjudge.Theycanalsomodifyprogramswithinacertainrangeaccordingtochangesinexternalconditions..Thedifferenceisthattheprincipleofmodifyingtheprogramisnotprescribedbyhumans,buttheprincipleofmodifyingtheprogrambytherobotitselfthroughlearningandsummingupexperience.Soitsintelligenceishigherthanthatofprimaryintelligentrobots.Thiskindofrobotalreadyhascertainautomaticplanningability,canarrangeitsownworkbyitself.Thiskindofrobotcanworkcompletelyindependentlywithouthumancare,soitiscalledadvancedself-disciplinerobot.Thiskindofrobotisalsobeginningtobecomepractical.
Keytechnology
Withtheneedsofsocialdevelopmentandtheexpansionofrobotapplicationfields,people'srequirementsforintelligentrobotsaregettinghigherandhigher.Theenvironmentinwhichintelligentrobotsarelocatedisoftenunknownandunpredictable.Intheprocessofresearchingsuchrobots,thefollowingkeytechnologiesaremainlyinvolved:
Multi-sensorinformationfusion
Multi-sensorinformationfusiontechnologyisaverypopularresearchtopicinrecentyears.Itiscombinedwithcontroltheory,signalprocessing,artificialintelligence,probabilityandstatisticstoproviderobotsinvariouscomplex,dynamic,uncertainandunknownPerformingtasksinanenvironmentprovidesatechnicalsolution.Therearemanykindsofsensorsusedinrobots,whicharedividedintotwocategories:internalmeasurementsensorsandexternalmeasurementsensorsaccordingtodifferentuses.Theinternalmeasurementsensorisusedtodetecttheinternalstateoftherobotcomponents,including:specificposition,anglesensor;arbitraryposition,anglesensor;speed,anglesensor;accelerationsensor;tiltanglesensor;azimuthanglesensor,etc.Externalsensorsinclude:vision(measurement,recognitionsensors),touch(contact,pressure,slidingsensors),force(force,torquesensors),proximity(proximity,distancesensors),andanglesensors(tilt,direction,Attitudesensor).Multi-sensorinformationfusionreferstotheintegrationofsensorydatafrommultiplesensorstoproducemorereliable,accurateorcomprehensiveinformation.Thefusedmulti-sensorsystemcanmoreperfectandaccuratelyreflectthecharacteristicsofthedetectedobject,eliminatetheuncertaintyofinformation,andimprovethereliabilityofinformation.Thefusedmulti-sensorinformationhasthefollowingcharacteristics:redundancy,complementarity,real-timeandlow-cost.Multi-sensorinformationfusionmethodsmainlyincludeBayesianestimation,Dempster-Shafertheory,Kalmanfilter,neuralnetwork,wavelettransform,etc.
Multi-sensorinformationfusiontechnologyisaveryactiveresearchfield.Themainresearchdirectionsare:Theweaknessofintegrity,sosingle-layerdatafusionlimitsthecapabilityandrobustnessofthesystem.Foradvancedsystemsthatrequirehighrobustnessandflexibility,amulti-levelsensorfusionmethodcanbeused.Low-levelfusionmethodscanfusemulti-sensordata;middle-levelfusionmethodscanfusedataandfeaturestoobtainfusedfeaturesordecisions;high-levelfusionmethodscanfusefeaturesanddecisionstothefinaldecision.
2MicrosensorsandsmartsensorsTheperformance,price,andreliabilityofsensorsareimportantindicatorstomeasurethequalityofsensors.However,manysensorswithgoodperformancelimittheapplicationmarketduetotheirlargesize.Therapiddevelopmentofmicroelectronicstechnologyhasmadeitpossibletomanufacturesmallandmicrosensors.Thesmartsensorintegratesthemainprocessing,hardwareandsoftware.Forexample,the1000seriesdigitalquartzsmartsensordevelopedbyParScientific,theolfactorysensordevelopedbyHitachiResearchInstituteofJapanthatcanidentify4kindsofgases,andtheDSTJ23000smartdifferentialpressuresensordevelopedbyHoneywelloftheUnitedStates,allhaveacertaindegreeofintelligence.
3Adaptivemulti-sensorfusionIntherealworld,itisdifficulttoobtainaccurateinformationabouttheenvironment,anditisimpossibletoensurethatthesensorcanalwaysworknormally.Therefore,forvariousuncertainsituations,arobustfusionalgorithmisnecessary.Someadaptivemulti-sensorfusionalgorithmshavebeendevelopedtodealwiththeuncertaintycausedbyimperfectsensors.Forexample,Hongproposedanextendedjointmethodthroughinnovativetechnology,whichcanestimatetheoptimalKalmangainofasinglemeasurementsequencefilter.PaciniandKoskoalsodevelopedanadaptivetargettrackingfuzzysystemthatcanbeappliedunderslightenvironmentalnoise.ItcombinestheKalmanfilteralgorithmintheprocessingprocess.
Navigationandpositioning
Inroboticssystems,autonomousnavigationisacoretechnologyandakeyanddifficultissueinthefieldofroboticsresearch.Thebasictasksofnavigationhavethreepoints:(1)Globalpositioningbasedonenvironmentalunderstanding:Throughtheunderstandingoftheenvironmentintheenvironment,identifyhumanroadsignsorspecificobjectstocompletethepositioningoftherobotandprovidematerialsforpathplanning;(2)GoalsRecognitionandobstacledetection:Real-timedetectionandrecognitionofobstaclesorspecifictargetstoimprovethestabilityofthecontrolsystem;(3)Safetyprotection:ItcananalyzeobstaclesandmovingobjectsintherobotworkingenvironmentandavoidcausingdamagetotherobotDamage.
Robotshaveavarietyofnavigationmethods.Accordingtodifferentfactorssuchasthecompletenessofenvironmentalinformationandthetypeofnavigationindicationsignals,theycanbedividedintothreetypes:map-basednavigation,map-basednavigation,andmap-freenavigation..Accordingtothedifferenthardwareusedinnavigation,thenavigationsystemcanbedividedintovisualnavigationandnon-visualsensorintegratednavigation[8].Visualnavigationusesacameratodetectandidentifytheenvironmenttoobtainmostoftheinformationinthescene.Thecontentofvisualnavigationinformationprocessingmainlyincludes:compressionandfilteringofvisualinformation,roaddetectionandobstacledetection,identificationofspecificenvironmentalsigns,andthree-dimensionalinformationperceptionandprocessing.Non-visualsensornavigationreferstotheuseofmultiplesensorstoworktogether,suchasprobetype,capacitivetype,inductivetype,mechanicalsensor,radarsensor,photoelectricsensor,etc.Thestatus,etc.aremonitoredtoperceivethestaticanddynamicinformationoftheworkingenvironmentoftherobot,sothatthecorrespondingworksequenceandoperationcontentoftherobotcannaturallyadapttothechangesoftheworkingenvironment,andeffectivelyobtaininternalandexternalinformation.
Inautonomousmobilerobotnavigation,whetheritislocalreal-timeobstacleavoidanceorglobalplanning,itisnecessarytoaccuratelyknowthecurrentstateandpositionoftherobotorobstacletocompletetaskssuchasnavigation,obstacleavoidanceandpathplanning.Thisisthepositioningoftherobot.Morematurepositioningsystemscanbedividedintopassivesensorsystemsandactivesensorsystems.Thepassivesensorsystemsensestherobot'sownmotionstatethroughcodediscs,accelerationsensors,gyroscopes,Dopplervelocitysensors,etc.,andobtainspositioninginformationthroughcumulativecalculations.Theactivesensorsystemsensestherobot'sexternalenvironmentorartificiallysetroadsignsthroughactivesensorsincludingultrasonicsensors,infraredsensors,laserrangefinders,andvideocameras,andmatchesthepre-setmodelofthesystemtoobtainthecurrentrobotandenvironmentorTherelativepositionoftheroadsignstoobtainpositioninginformation.
Pathplanning
Pathplanningtechnologyisanimportantbranchinthefieldofrobotics.Theoptimalpathplanningisbasedononeorsomeoptimizationcriteria(suchasthesmallestworkcost,theshortestwalkingroute,theshortestwalkingtime,etc.),tofindapathfromthestartingstatetothetargetstateintherobotworkspace,whichcanavoidobstaclesTheoptimalpath.
Pathplanningmethodscanberoughlydividedintotwotypes:traditionalmethodsandintelligentmethods.Traditionalpathplanningmethodsmainlyincludethefollowingmethods:freespacemethod,graphsearchmethod,griddecouplingmethod,andartificialpotentialfieldmethod.Mostoftheglobalplanninginrobotpathplanningisbasedontheabove-mentionedmethods,butthesemethodsneedtobefurtherimprovedintermsofpathsearchefficiencyandpathoptimization.Theartificialpotentialfieldmethodisamorematureandefficientplanningmethodintraditionalalgorithms.Itusestheenvironmentalpotentialfieldmodeltoplanthepath,butitdoesnotinvestigatewhetherthepathisoptimal.
Theintelligentpathplanningmethodistoapplyartificialintelligencemethodssuchasgeneticalgorithm,fuzzylogicandneuralnetworktopathplanningtoimprovetheobstacleavoidanceaccuracyofrobotpathplanning,acceleratetheplanningspeed,andmeettheneedsofpracticalapplications.Amongthem,themostusedalgorithmsmainlyincludefuzzymethods,neuralnetworks,geneticalgorithms,Qlearningandhybridalgorithms,etc.Thesemethodshaveachievedcertainresearchresultswhentheobstacleenvironmentisknownorunknown.
RobotVision
Thevisionsystemisanimportantpartofautonomousrobots,generallycomposedofcameras,imagecapturecardsandcomputers.Theworkoftherobotvisionsystemincludesimageacquisition,imageprocessingandanalysis,outputanddisplay.Thecoretasksarefeatureextraction,imagesegmentationandimagerecognition.Howtoprocessvisualinformationaccuratelyandefficientlyisthekeyissueofthevisualsystem.Visualinformationprocessingisgraduallyrefined,includingcompressionandfilteringofvisualinformation,environmentandobstacledetection,identificationofspecificenvironmentalsigns,andthree-dimensionalinformationperceptionandprocessing.Amongthem,environmentandobstacledetectionisthemostimportantanddifficultprocessinvisualinformationprocessing.Edgeextractionisacommonlyusedmethodinvisualinformationprocessing.Forgeneralimageedgeextraction,suchasthegradientmethodoflocaldataandthesecond-orderdifferentiationmethod,itisdifficulttomeetthereal-timerequirementsformobilerobotsthatneedtoprocessimagesinmotion.Forthisreason,amethodofimageedgeextractionbasedoncomputationalintelligencehasbeenproposed,suchasaneuralnetwork-basedmethodandamethodusingfuzzyinferencerules.Inparticular,ProfessorBezdekJ.Chasrecentlycomprehensivelydiscussedtheuseoffuzzylogicinferenceforimageedgeextraction.significance.Thismethodisspecifictovisualnavigation,whichistointegratetheroadknowledgerequiredbytherobotwhenmovingoutdoors,suchasroadwhitelinesandroadedgeinformation,intothefuzzyrulelibrarytoimprovetheefficiencyandrobustnessofroadrecognition.Othershaveproposedtocombinegeneticalgorithmswithfuzzylogic.
Robotvisionisoneofthemostimportantsignsofitsintelligence,whichisofgreatsignificancetorobotintelligenceandcontrol.Vigorousresearchisbeingcarriedoutathomeandabroad,andsomesystemshavealreadybeenputintouse.
Intelligentcontrol
Withthedevelopmentofrobotics,thetraditionalcontroltheoryexposesphysicalobjectsthatcannotbeaccuratelymodeledandpathologicalprocesseswithinsufficientinformation.Duetoshortcomings,manyscholarshaveproposedvariousintelligentrobotcontrolsystemsinrecentyears.Theintelligentcontrolmethodsofrobotsincludethefusionoffuzzycontrol,neuralnetworkcontrol,andintelligentcontroltechnology(thefusionoffuzzycontrolandvariablestructurecontrol;thefusionofneuralnetworkandvariablestructurecontrol;thefusionoffuzzycontrolandneuralnetworkcontrol;thefusionofintelligentfusiontechnologyisalsoIncludingfuzzycontrolmethodsbasedongeneticalgorithms),etc.
Robotintelligentcontrolhasmadegreatprogressintheoryandapplication.Intermsoffuzzycontrol,J.J.Buckleyetal.demonstratedtheapproximationcharacteristicsofthefuzzysystem,andE.H.Mamdanappliedfuzzytheorytoanactualrobotforthefirsttime.Fuzzysystemshavebeenwidelyusedinvariousfieldssuchasrobotmodeling,control,flexiblearmcontrol,fuzzycompensationcontrol,andmobilerobotpathplanning.Intermsofrobotneuralnetworkcontrol,CMCA(Cere-bellaModelControllerArticulation)isanearliercontrolmethod.Itsbiggestfeatureisstrongreal-time,especiallysuitableforthecontrolofmulti-degree-of-freedommanipulators.
Theintelligentcontrolmethodimprovesthespeedandaccuracyoftherobot,butitalsohasitsownlimitations.Verysimple,theaccuracyofcontrolwillbelimited;whetheritisfuzzycontrolorvariablestructurecontrol,chatteringwillexist,whichwillhaveaseriousimpactonthecontrol;thenumberofhiddenlayersoftheneuralnetworkandthenumberofneuronsinthehiddenlayerThereasonabledeterminationofisstillaproblemencounteredbyneuralnetworksincontrol.Inaddition,neuralnetworksareeasilytrappedinlocalminimumsandotherproblems,whichareproblemstobesolvedinintelligentcontroldesign.
Human-machineinterfacetechnology
Theresearchgoalofintelligentrobotsisnottocompletelyreplacehumans.Itisdifficultforcomplexintelligentrobotsystemstorelysolelyoncomputerstocontrolthem.Yes,evenifitcanbedone,itisnotpracticalduetolackofadaptabilitytotheenvironment.Intelligentrobotsystemscannotcompletelyexcludetheroleofhumans,butneedtorelyonhuman-machinecoordinationtoachievesystemcontrol.Therefore,awell-designedhuman-machineinterfacehasbecomeoneofthekeyissuesintheresearchofintelligentrobots.
Human-machineinterfacetechnologyisthestudyofhowtomakepeoplecommunicatewithcomputerseasilyandnaturally.Inordertoachievethisgoal,inadditiontothemostbasicrequirementthattherobotcontrollerhasafriendly,flexibleandconvenientman-machineinterface,thecomputerisalsorequiredtobeabletoreadtext,understandlanguages,speakexpressions,andevenbeabletocommunicateindifferentlanguages.Therealizationofthesefunctionsdependsontheresearchofknowledgerepresentationmethods.Therefore,thestudyofhuman-computerinterfacetechnologyhasbothgreatapplicationvalueandbasictheoreticalsignificance.Human-machineinterfacetechnologyhasachievedremarkableresults,andtechnologiessuchastextrecognition,speechsynthesisandrecognition,imagerecognitionandprocessing,andmachinetranslationhavebeguntobepractical.Inaddition,human-machineinterfacedevicesandinteractiontechnology,monitoringtechnology,remoteoperationtechnology,communicationtechnology,etc.arealsoimportantcomponentsofhuman-machineinterfacetechnology,amongwhichremoteoperationtechnologyisanimportantresearchdirection.
Developmentdirection
Althoughroboticartificialintelligencehasachievedremarkableresults,cyberneticsexpertsbelievethatthelimitofthelevelofintelligenceitcanpossesshasnotbeenreached.Theproblemisnotonlytheinsufficientcomputingspeedofthecomputerandthefewtypesofsensorysensors,butalsootheraspects,suchasthelackofdesignideasforprogrammingtherationalbehaviorofrobots.Doyouthinkthateventhethinkingprocessofhumanbeingswhensolvingthemostcommonproblemsisnotdeciphered.Whatabouthumanintelligence?Thiskindofunderstandingprocessisprogressingveryslowly.Howcanwemasterthelawsandmakethecomputer"think"faster??Therefore,theproblemofnotknowinghumansthemselveshasbecomeastumblingblockinthedevelopmentofrobots.Thesubjectofmakingintelligentrobotsthat"live"inanunstableenvironmenthasinrecentyearsenabledpeopletoconductin-depthresearchontheprocessesofcognitionandself-recognitionthatoccurinbiologicalsystems,animals,andhumanbrains.Theresultisthetheoryofhierarchicaladaptivesystems,whichisbeingeffectivelydeveloped.Asthetheoreticalbasisfororganizingintelligentrobotstoperformpurposefulbehaviors,howdoesourbraincontrolourbodies?Roughlyestimatedpurelyfromamechanicalpointofview,ourbodyalsohasmorethantwohundreddegreesoffreedom.Whenwearedoingcomplexactionssuchaswriting,walking,running,swimming,andplayingthepiano,howdoesthebraingiveorderstoeverymuscle?Howcanthebrainprocesssomuchinformationintheshortesttime?Ourbrainisnotinvolvedintheseactivitiesatall.Thebrain-ourcentralinformationprocessingmachine"disdains"totakecareofthis.Itdoesnotsupervisethevariousmovementpartsofourbodyatall.Thedetaileddesignofthemovementsiscarriedoutatamuchlowerlevelthanthecerebralcortex.Thisismuchlikeprogramminginahigh-levellanguage.Aslongasyoupointout"asetofnumbersfrom1to20withanintervalofone",therobotitselfwillinputthissetofinstructionsintoadetailedoperatingsystem.Themostobviousisthatthemostobviousinstructionssuchas"retractyourhandsassoonasyoutouchahotobject"areissuedevenbeforethebrainisawareofit.
Distributingalargetaskamongseveralcorticallayersismorecost-effective,economical,andeffectivethanstrictlycentralizeddistributioninwhichthecontrolorgansstipulatethenecessaryactionsforeachelementofthesystem.Whensolvingmajorproblems,suchacentralizedbrainwillappeartoocomplicated,notonlythebrain,buteventheentirehumanbodycannotaccommodate.Whencompletingsomecomplexmovementsofonekindoranother,weusuallybreakthemdownintoaseriesofcommonsmallmovements(suchasgettingup,sittingdown,steppingrightfoot,steppingleftfoot).Thevariousactionstaughttochildrencanbeattributedtotheformationandconsolidationofcorrespondingsmallactionsinthechild's"memory".Inthesameway,theperceptionprocessisorganizedinthisway.Perceptualimage-thisisafixedsequenceorcombinationofauditory,visualortactileimpulses(horse,human),oracombinationofboth.Learningabilityisanothergeneralprincipleoforganizationalcontrolincomplexbiologicalsystems.Itistheabilitytoadapttothepreviouslyunknownandchanginglivingenvironmentinafairlywiderange.Thisadaptabilityisnotonlyinherenttotheentirebody,butalsoinherenttoindividualorgansandevenfunctionsofthebody.Thisabilityisirreplaceablewhenthesameproblemshouldbesolvedmultipletimes.Itcanbeseenthatthephenomenonofadaptabilityplaysanextremelyimportantroleinthepurposefulbehavioroftheentirebiologicalworld.
Theproblemofcontrollingrobotsistosimulateanimalmovementsandhumanadaptability.Establishthelevelofrobotcontrol-thefirstistoimplementthedistributionofperceptionfunctions,informationprocessingfunctions,andcontrolfunctionsateachleveloftherobotandbetweensubsystems.Thethird-generationrobothaslarge-scaleprocessingcapabilities.Inthiscase,thecompletelyunifiedalgorithmforinformationprocessingandcontrolisactuallyinefficientorevenuseless.Therefore,theemergenceofthehierarchicaladaptivestructureisfirsttoimprovethequalityofrobotcontrol,thatis,toreducethelevelofuncertaintyandincreasethespeedofaction.Inordertoplaytheroleofvariouslevelsandsubsystems,theamountofinformationmustbegreatlyreduced.Therefore,thevariousfunctionsofthealgorithmallowpeopletocompletetaskswithgreatlyreduceduncertainty.Inshort,thedevelopmentofintelligenceisanimportantfeatureofthethirdgenerationofrobots.Peopledecidewhichrobotgenerationtheybelongtoaccordingtotheintelligenceleveloftherobot.Somepeopleevendividerobotsintothefollowingcategoriesaccordingly:Controlledrobots-"zero-generation"robots,whichdonotpossessanyintellectualperformanceandaremanipulatorscontrolledandmanipulatedbyhumans;robotsthatcanbetrained-thefirst-generationrobots,Hasamemory,isoperatedbyhumans,andtheactionplanandprogramarespecifiedbyhumans.Itjustremembers(theabilitytoreceivetraining)andreproducesit;feeltherobot-aftertherobotrememberstheplanarrangedbythehuman,itthendependsontheoutsideworld.Thedata(feedback)calculatesthespecificprogramoftheaction;theintelligentrobot-afterthehumanspecifiesthetarget,therobotindependentlycompilestheoperationplan,determinestheactionprogramaccordingtotheactualsituation,andthenchangestheactionintothemotionoftheoperatingmechanism.Therefore,ithasawiderangeofsensorysystems,intelligence,andsimulationdevices(thesurroundingsituationanditself-therobot'sconsciousnessandself-awareness)
Researchfocus
Howtobecomesmart
Artificialintelligenceexpertspointedout:Computersshouldnotonlydowhathumansspecifytodo,buttheyshouldalsosolvemanythingsinthebestwayontheirown.Forexample,theentireprogramofanordinarycomputerthatcalculateselectricitybillsorisengagedinbankingbusinessistocompletetheinstructionlistaccurately,whilethecomputerofsomescientificresearchcenterswill"think"theproblem.Theformerrunsfast,butitisnotintelligent;thelatterstoresmorecomplicatedprograms,andthecomputerisfilledwithinformation,whichcanimitatemanyhumanabilities(insomecasesevenmorethanourhumanabilities).
Moreintelligence
Scientistsbelievethattheresearchanddevelopmentdirectionofintelligentrobotsistoinstall"brainchips"onrobotstomakethemmoreintelligent.Automatedorganization,comprehensiveprocessingoffuzzyinformation,etc.willtakeabigstepforward.
Althoughsomepeoplehaveexpressedconcern:Willthisintelligentrobotequippedwitha"brainchip"surpasshumansinintelligenceinthefuture,andevenposeathreattohumans?Butmanyscientistsbelievethatsuchworriesarecompletelyunnecessary.Asfarasintelligenceisconcerned,theIQofarobotisequivalenttothatofa4-year-oldchild,andthe"commonsense"ofarobotisfarworsethanthatofanormaladult.
NationalPolicy
TheMinistryofIndustryandInformationTechnology,theNationalDevelopmentandReformCommission,andtheMinistryofFinancejointlyissuedthe"RobotIndustryDevelopmentPlan(2016-2020)",pointingoutthatrobotsIndustrialdevelopmentmustpromotemajorlandmarkproductstotaketheleadinmakingbreakthroughs.
Inthefieldofindustrialrobots,focusonintelligentproductionandintelligentlogistics,overcomekeytechnologiesofindustrialrobots,improveoperabilityandmaintainability,andfocusonthedevelopmentofarcweldingrobots,vacuum(clean)robots,andcompleteSixiconicindustrialrobotproducts,includingintelligentindustrialrobots,human-machinecollaborativerobots,dual-armrobots,andheavy-dutyAGVs,havebeenindependentlyprogrammedtoguidethedevelopmentofmycountry'sindustrialrobotstothemid-to-highend.
Inthefieldofservicerobots,focusonthedevelopmentof4iconicproductssuchasfirerescuerobots,surgicalrobots,intelligentpublicservicerobots,andintelligentnursingrobots,andpromotetherealizationofprofessionalservicerobots.Thecommercializationofpersonal/homeservicerobotsisrealized.
Thecountryhasformulatedcertainstandardsforthetechnologies,specificationsandfunctionsoftheaboveteniconicproducts.Forexample:intelligentpublicservicerobots.Navigationmethod:laserSLAM,maximummovingspeed0.6m/s,positioningaccuracy±100mm,positioningheadingangleaccuracy±5°,maximumworkingtime3h,numberofarms2,singlearmdegreesoffreedom2-7,headdegreesoffreedom1-2.Ithasthefunctionsofautonomouswalking,human-computerinteraction,explanationandguidance.
Summary
Intelligentrobots,asatechnologythatcontainsconsiderableknowledgeofmultipledisciplines,arealmostproducedwithartificialintelligence.Intelligentrobotsarebecomingmoreandmoreimportantintoday'ssociety,andmoreandmorefieldsandpositionsrequiretheparticipationofintelligentrobots,whichmakestheresearchofintelligentrobotsmoreandmorefrequent.Althoughitisstilldifficultforustoseetheshadowofintelligentrobotsinourlives.Butinthenearfuture,withthecontinuousdevelopmentandmaturityofintelligentrobottechnology.Withtheunremittingeffortsofmanyscientificresearchers,intelligentrobotswillsurelyenterthousandsofhouseholds.Tobetterservepeople’slivesandmakepeople’slivesmorecomfortableandhealthy.