(ROBUST) CONTROL THEORY FUNDAMENTALS
Transcript of (ROBUST) CONTROL THEORY FUNDAMENTALS
(ROBUST)CONTROLTHEORYFUNDAMENTALS PabloA.Parrilo-MITEECS6.S979TopicsinDeployableMLFall2019
AroundRobustness… • RobustStatistics(1960s--)• RobustControl(1980s--)• RobustOptimization(1990s--)• RobustEconomics(2000s--)• RobustMachineLearning(2015s--)
Historyrepeatsitself,firstastragedy,secondasfarce.
KarlMarx
Thislecture
• Whatis“ControlTheory”?• Whatis“RobustControl”?• (Some)insightsandresults• Afewtechnicalnuggets• Modelsofuncertainty,andassessmentofrobustness• ManyquestionsthatIdon’tknowtheanswerto
DynamicsandControlTwokeynotions:Dynamicalsystems:• Systemsthatevolveover“time”Control/Decisions:• Valuesofsomeofthevariablescanchosenbyus.
Manydifferentflavors…• Continuous/DiscreteTime• Differential/DifferenceEquations• Differential/AlgebraicEquations(DAEs)• StochasticDifferentialEquations(SDEs)• MarkovDecisionProcesses(MDPs)• DiscreteEventSystems(DES)• (more…)Also:• Single/MultipleDecisionMakers(e.g.,games)• “Classical”vs.“Adaptive”vs.“Robust”
Aroughclassification
OneDecisionMaker
SeveralDecisionMakers
Static(notime/sequential)
Optimization GametheoryRobustOptimization
Dynamic(sequentialdecisions)
OptimalControl DynamicGamesRobustControl
Keynotions
Analysis:Givenaclosed-loopsystem(G,K)andspecifications,determineifthespecificationsaresatisfied.Synthesis:Givenanopen-loopsystemGandspecifications,findacontrollerKsuchthatthespecificationsaresatisfied.
But…
Mathematical/Computationalmodels(regardlessofdetail)arealwaysmerelyapproximationstothe“real”phenomenon.Robustness:Simultaneouslymodelthesystem,andtheuncertaintyassociatedwithitE.g.,unknownparameters,uncertaindynamics,environment…
Optimalitydoesnotsuffice J.C.Doyle(1978),IEEETransactionsinAutomaticControl
State-of-the-artdesignmethods(optimalquadraticcontrol,LQG)Arbitrarilysmallperturbationscandestabilizesystem!
ModelsystemanduncertaintyManydescriptionsofuncertainty:
• Set-based:unknownelementsinagivenset(worst-case)
• Probabilistic:Setofallowableperturbations,plusadistribution
Whatcanweefficientlydo?Models/theoremsaregreat,butwhentherubbermeetstheroad,whatreallymattersiswhatyoucancomputeeffectively(andthismaymeandifferentthingstodifferentpeople)Formulatingvs.solvingOptimalControl:Hamilton-Jacobi-Bellman(HJB)equationNonlinearfiltering:Kushner,ZakaiequationsDynamicProgramming:“inprinciple”cansolveanywell-definedgame,but..
(Ontheboard…)Linearsystems–whydolikethemsomuch,andwhatwecandowiththemGoingnonlinear–analysisisgreat,synthesisnotsomuchUncertainty–IQCmodelsBottomline:incertaincases,reliable,algorithmicanalysis/designmethods
AFewLessons“Optimality”doesnotguaranteerobustness–infact,mayhurtitValuablestructuralinsights(e.g.,controllability/observability)Cleverreformulationscanhelpalot,evenifintheendwestillresorttocomputationalmethodsValuabletoolkittoexploitstructure
Certifyingperformance
• Semidefiniterelaxationsarecentralintoday’srobustcontrol/optimization
• Why?Becausewedon’tknowmanywaysofrigorouslycertifyingpropertiesoffunctionsinmanyvariables.
• Speculative:linksbetweenrobustnessandverifiability
BillionDollarquestion
WhatwillbetheepistemologyofdeployableML?Howwillweconvinceourselvesthatthings“work”?Whatwillthismean?Theorems?Extensivesimulations?Distinctionbetweencat/no-catvs.safety-critical?(e.g.,90000dailyflightsinUS.99.9%successrate->90dailyplanecrashes)Howtobridgeacrossallthesedifferentcultures?