行为基础机器人编程：入门与实践指南

需积分: 7 158 浏览量更新于2024-07-18 收藏 6.13MB PDF 举报

"《行为机器编程：APracticalGuidetoBehavior-BasedRobotics》是一本由Joseph L. Jones编著的专业书籍，专为对智能机器人领域感兴趣的人提供了一个实用且易于理解的入门指南。该书深入浅出地探讨了如何通过行为驱动的方法来编程机器人，适合未来希望在这个快速发展的行业中发展的读者。作者在书中详细阐述了行为基础的机器人设计原理，包括如何理解和模拟机器人在复杂环境中的行为决策过程，以及如何利用诸如传感器输入、运动控制和学习算法等核心组件构建智能行为。读者可以通过这本书了解到如何将高级任务分解为一系列可执行的行为模块，从而使机器人能够适应多变的场景并自主解决问题。《行为机器编程》还可能结合了Daniel Roth的Robotic Simulator作为实践工具，读者可以借助这个模拟器进行实验和测试，提升理论知识与实践技能的结合。此外，该书特别强调版权保护，所有复制或分发必须得到McGraw-Hill公司的授权。本书不仅适用于课堂教学，也适合独立学习者使用，它涵盖了从基础概念到高级技术的广泛内容，包括但不限于行为规划、控制策略、以及机器学习在行为识别和决策制定中的应用。对于那些希望通过行为机器编程技术推动机器人技术进步的人来说，这是一本不可或缺的参考资料。《行为机器编程》是一本实用性强、理论与实践相结合的教材，旨在帮助读者掌握行为驱动的机器人编程方法，为进入智能机器人行业奠定坚实的基础。无论是希望成为机器人工程师、科研人员还是开发者，这本书都将提供宝贵的知识和实践经验。"

Introduction

Therearemanywaystoprogramamobilerobot.Theleastcomplexrobotshave

programswritteninsolder

wheresensorsareconnectedmoreorlessdirectlyto

motors.Attheotherextremearerobotsprogrammedinhigh-levellanguages—

languageslikeLispandJavathatareoftenusedtosupportartificialintelligence.But

regardlessofhowtheprogrammingisaccomplished,allrobotprogramsexhibitsome

sortofstructureorarchitecture.

Programscomposedbybeginningroboticistsoftenhaveastructurethatmightbe

describedasadhocorperhapsorganic—theprogramsjustgrow.Withoutany

overarchingprinciplesormethodology,theprogrammerthinksofafeatureandwrites

thecodethatimplementsthatfeature.Asecondfeaturespringsfromthefertilemind

oftheprogrammerandiscombinedwiththefirst,athirdfeaturecomestobe,andso

on.Asdevelopmentproceeds,theorganicanalogybecomesmoreandmoreapropos.A

bitofcodethatimplementsoneaspectoftheprogramintertwineswithcodethat

implementsanother,magicnumberstakeroot,andspecialcasesflourishandspread.

Theorganicapproachdoeswork.Withenoughtime,patience,andcodespace,the

programmercancoerceanorganicallystructuredrobotprogramintoproducing

desiredresults.Butwithouttheguidanceofrobot-specificprinciples,suchprograms

becomemoreandmoreconvoluted.Italwaysseemstotakelongertoimplementthe

nextfeaturethanittooktoimplementthelast;programmersoftenrefrainfrom

uprootingobsoletefeaturesbecausebugstendtocropupwhen“unused,”buthighly

cross-coupledsectionsofcodearedeleted.Evenwhenprogrammersfollowmodern

tenetsofmodularityandothergoodprogrammingpractices,robotprogramscan

becomecumbersomeandfragile.

Robot/ComputerDifferences

Programmingtroubledevelopsbecausebeginningroboticistsoftenmissthe

importanceofabasicfact:robotsandcomputersaredifferent.Thegoalsofarobot

programandacomputerprogramaredifferent;theconstraintsimposedoncomputers

androbotsaredifferent—andthesedifferencesarecrucial.Aroboticistmust

understandandrespectthedistinctionsifheorsheistodevelopeffectiverobot

programs.

SerialversusParallel

Theintuitionthatprogrammersdeveloplearningtoprogramcomputersoftendoesnot

carryovertoprogrammingrobots.Oneaspectofprogrammerintuitionthatsometimes

failsconcernstheissueofserialversusparallelexecution.Serialexecutionis

satisfactoryformostcomputerprograms,butrobotprogramsdemandaparallel

approach.

Atypicalcomputerprogramisdesignedtocomputeananswerandreturnaresult.At

theircore,evenhighlyinteractiveandintenselygraphicalcomputerprogramslike

videogamesfollowthisbasicapproach.Computationproceedsinasequenceofsteps

where,typically,theoutputofonestepbecomestheinputofthenextstep.Thetotal

timerequiredtoreachanansweristhesumofthetimestakenforeachstep.Therefore,

afastercomputerisalwaysabettercomputer—theuserofafastercomputergetsthe

answermorequicklythantheuserofaslowercomputer.

Alternately,afastercomputercangiveamorepreciseanswerorcangivemore

answersinthesametimetakenbyaslowcomputer.Foravideogame,computerspeed

translatesdirectlyintobetterresolutionandmorerealisticsimulations.

Butcomputingananswerisnotthepurposeofanautonomousmobilerobot.Rather,a

robotseekstoachieveagoalormaintainastatewhileavoidinghazardsandtraps—

verymuchasalivingorganismdoes.Therobotmustattendsimultaneouslytoallofits

concerns;e.g.,don’tcollidewithanything,don’tfalldownthesteps,don’trunoutof

powerfarfromthecharger.Catastrophemightresultif,forexample,therobotwereto

neglectwatchingoutfortheedgeofthestairswhileconcentratingexclusivelyon

avoidingacollisionwiththebaseboard.Arobotcrashcanbearathermoreserious

matterthanacomputercrash.

PlansversusOpportunities

Atypicalcomputerprogramexecutesaplan—onethinghappensafteranotheruntila

resultisreached.Butanautonomousrobotneedstobeopportunistic—sometimesthe

robot’sgoalisalreadyachieved;alltherobothastodoisnotice.Forexample,suppose

wewantourrobottofinditswaytoachargerwhenthebatteriesarelow.Aplan-based

programdecidingtorechargethebatteriesmightfirsthavetherobotsearchfora

centralbeaconintheroom,gotothebeacon,orientitselfinaparticulardirection

(towardthecharger),thenproceeduntiltherobotencountersthelocalbeaconthat

indicatesthecharger.

Thissoundslikeareasonableplan—unlesstherobothappenstobepositionedonlya

footfromthechargerwhentherobotdecidesthatthebatteriesneedrecharging.Inthat

case,theplan-followingrobotfirstmovesawayfromthechargertofindthecentral

beacon,onlytoturnbackimmediatelytowardthecharger.Amoresensibleapproach

wouldhavetherobotnoticethatitwasnexttothechargertobeginwith,andthenact

accordingly,ratherthanfollowaplanblindly,ignoringreal-worldopportunities.

Constantinputfromsensorsandaprogrammingparadigmabletomakeuseoftherush

ofdataareneededtoenablearobottotakeadvantageofitsopportunities.

GracefulDegradation

Givenaccuratedataandalogicallycorrectprogram,acomputerwillreturnacorrect

result.Giveninaccuratedata,thecomputer’soutputisunreliable.Thisobservation

givesrisetothewell-knownGIGOadage,“Garbagein,garbageout.”Thecomputer

dependsonahumanoperatortoinputaccuratedataandhaslittlerecourseifthedata

arewrong.Thusincorrectcomputerresponsesareascribedto“humanerror.”

Anautonomousrobotcollectsitsowndatathroughitssensors.Andsensors,aswe

shallsee,oftenmislead.Thelessyoupayforasensor,thelessinclineditistoward

veracity.Butevenveryexpensivesensorsprovideunreliabledataincommon

situations.Becauseoftheunreliabilityofitsinputs,arobotprogrammustbe

engineeredinsuchawaythattherobotworksaswellaspossibleunderthe

circumstances.Thatis,robotperformanceshoulddegradegracefullyinthepresenceof

inaccurateormissingdata.Arobotprogrammustnotcollapseinaheap(asa

computerprogrammight)atthefirstsignoferroneousinput.

Behavior-BasedAdvantages

Behavior-basedapproachestorobotprogrammingexcelatparallelexecution,

opportunisticgoalrealization,andgracefuldegradation.Programmingarobot

accordingtobehavior-basedprinciplesmakestheprograminherentlyparallel,

enablingtherobottoattendsimultaneouslytoallthehazardsitmayfaceaswellasthe

serendipitousopportunitiesitmayencounter.Further,behavior-basedrobotscaneasily

accommodatemethodsthatallowperformancetodegradegracefullyinthepresenceof

sensorerrororfailure.

What’sAlltheFuss?

Experiencedprogrammersmaybewondering:Isthereanythingnewhere?Isn’tan

autonomousmobilerobotreallyjustanexampleofanembeddedsystemplusareal-

timeoperatingsystem?Might“behavior-basedprogramming”benothingmorethana

fancynameappliedtocommonplacepractices?

Mobilerobotsdoindeedqualifyasembeddedsystems.Andwhethertherobot’s

softwarerunsunderacommerciallysuppliedreal-timekernelorisimplementedin

rawcode,everyrobotmusthavesomesemblanceofareal-timeoperatingsystem.

Further,thegoodpracticesaroboticistfollowsinconstructingabehavior-based

programarenotinconsistentwiththegoodpracticesanembeddedsystemprogrammer

mightfollowwritingcodethatcontrols,say,aDVDplayeroracellphone.

However,bythemselves,commonpracticesforwritingembeddedsystemcodearenot

sufficientforconstructingeffectiverobotprograms.Autonomousrobots,aswill

becomeincreasinglyclear,regularlyconfrontchallengesrarelyfacedbyother

embeddedsystems.Dealingwiththesechallengescallsfortheadditionalsetof

organizingprinciplesthatbehavior-basedprogrammingsupplies.

Focus

Thebehavior-basedrobotprogrammingparadigmisaneminentlypracticaloneand

likewise,throughoutthistext,mytreatmentofbehavior-basedroboticswillfocus

moreonpracticalissuesthanonacademicrigor.

Myintentionistoprovideyouwith

importantfundamentalsthatwillhelpyouprogramrobotseffectively.Ihopethatyou

willtakeawayfromyourstudyofbehavior-basedroboticsanappreciationofthe

powerandscopeofthisrobotprogrammingmethodandthatyouwillfeelconfident

implementingabehavior-basedapproachinyourfutureroboticprojects.

Inthistext,Iwillexplainrobotprogrammingtheory,offerexamplesofsuccessful

robotprograms,andrelateinsightsthatIhavefounduseful.Butultimatelyyoucannot

learnrobotprogrammingfromabook.Tolearnrobotprogramming,youmust

programrobots.Unfortunately,becauseofcostandothercomplications,noteveryone

hasimmediateaccesstoaprogrammablemobilerobot.Toaddressthisdifficulty,my

colleague,DanielRoth,hasdevelopedaroboticsimulator,BSim;wehaveintegrated

BSimwiththetext.ToaccessBSimpleasevisittheWebsite:

www.behaviorbasedprogramming.com.

WerecommendthatyoumakeapracticeofrunningBSimfrequentlyasyouwork

towardunderstandingtheconceptspresentedhere.Asapedagogicaltool,robot

simulatorscanbequitehelpful.BSimletsyouisolateaspectsofrobotbehaviorand

slowdownprocessessothatyoucaneasilyobservewhatisgoingon.However,asa

predictorofhowaphysicalrobotwillinteractwithphysicalobjects,simulatorsare

notoriouslyunreliable.Donotassumethatwhatworksinsimulationwillworkinthe

realworld.Tofullyappreciatehowarobotbehavesinthephysicalworldthereisno

substituteforprogramminganactual,physicalrobot.

3–6

Prerequisites

Whatdoyouneedtoknowbeforeyoubegin?Becauserobotshavesuchabroad

appealandattractadiversesetofenthusiasts,itishardtochooseafirmsetof

prerequisiteknowledge.Eachreaderwillbringadifferentsetofinterests,experience,

andwillingnesstoacquiremissingunderstanding.However,ingeneral,areaderwill

haveaneasiertimeifheorshehasaworkingknowledgeofalgebraandtrigonometry

andhassomeexperiencewithvectors.Youwillmisslittleifyouhaveyettomaster

linearalgebraandcalculus.Itwillbehelpfultohavebasicfamiliaritywithcomputer

programmingbeforereadingthisbook,butnogreatdepthofexperienceisassumed.

Organization

InChapter1,weuseBSimtoobservethebehaviorofaworkingsystem,asimulated

robot.Thesimulatedrobotexhibitsinterestingandcomplexinteractionswithits

environment.Tounderstandtheseinteractionswetakeastepbackandask:what

exactlyisarobotandwhatareitsessentialcomponents?

InChapter2,wereviewthefeedbackcontrolsystem,atraditionalmethodof

connectingsensingtoactuation.ReturningtoBSim,weobservethefunctioningof

elementarycontrolsystems.Butfromtheverybeginningwewilllearnthepainful

lessonsofhowgoodcontrolsystemscangobad.

InChapter3,webegintobuildprimitivebehaviors,learningabouttriggersandabout

ballisticversusservobehaviors.

Chapter4dealswitharbiters,thesoftwareconstructthatallbehavior-basedsystems

musthavetomanageconflictsbetweenbehaviors.

Puttingtogetherallthatwehavelearned,inChapter5,wecanwritecomplete

programs.Wecommencefillingabagofusefulrobotprogrammingtricks.

Withthebasicsciencenowmastered,weproceedtotheartofrobotprogrammingin

Chapter6.Hereweaddresshowaproblemstatementcanbeconvertedintoarobot

program.Principlesandheuristicstoguidethisperilousbutnecessarystepare

suggested.

InChapter7,wetakeasoftware-centeredlookatsomecommonsensors,howthey

function,andthewaysthatsensoroutputcanbemisleading.

Chapter8recountsacasestudyofabehavior-basedrobotimplementation.This

implementationcanbeinstantiated

usingan8-bitmicrocontrollerusingtheC

computerlanguage.

RoboticsisavitalandgrowingfieldandinChapter9,wetakeabrieflookahead.A

fewofthesortsofrobotsthatmaysoonbepossiblearediscussed,asaresomeofthe

keymissingelementsholdingbackthearrivaloftheserobots.

Programmingarobotmeans,amongotherthings,specifyinghowarobotmoves.

AppendixAprovidessomeofthefrequentlyuseddetailsthataprogrammust

incorporatetocontrolrobotmotion.

AppendixBdescribesBSim,itsfunctioning,andconstruction.BSimisopen-source

softwarewritteninJava;youmaywishtoextendorspecializetheprogramforyour

ownpurposes.ConsulttheBSimWebsitefordetails.

InAppendixCareanumberofsimplefunctionsthatIhavefoundbeneficialinthe

constructionofmanydifferentrobotprograms.

Pseudocodeisusedthroughoutthetexttodescribetheworkingsofrobotprograms.

AppendixDdescribestheinformalformalismusedinthebook.

Inmyexperience,developingrobotsrequirestwothingsaboveallelse:technical

proficiencyandunfetteredcreativity.Theexercisesattheendofeachchapterare

intendedtopromoteboth.Manyoftheexercisesseektoexpandyourproficiencyby

askingyoutocomputeavalueorsolveanequation.Manyotherschallengeyouto

thinkaboutsomething.Thelatterexercisesmayhavenoanswerormanyanswers—

theirpurposeistostimulatecreativethought.

Buildingandprogrammingarobotisverymuchanactofcreation—anundertaking

thatIcontinuetofindimmenselysatisfying.Ihopeyouwillenjoyyourendeavorsin

thefieldofroboticsasmuchasIhavemine.

剩余258页未读，继续阅读

soupbox

粉丝: 0
资源: 5

行为基础机器人编程：入门与实践指南

吴恩达机器学习matlab编程作业

coursera机器学习编程作业8

机器学习的编程练习.zip

基于Matlab编程的人体行为识别[Matlab编程].zip

用Python实现吴恩达机器学习编程作业.zip

编程开发-编程工具-Shielden通用机器码解码工具 v1.0.zip

斯坦福大学机器学习第九周编程作业

Buzz:专为机器人群设计的编程语言

遗传编程与机器学习应用

机器学习入门：显著式编程与非显著编程的理解

最新资源