USF Libraries
USF Digital Collections

Analysis of VTOL MAV use during rescue and recovery operations following Hurricane Katrina

MISSING IMAGE

Material Information

Title:
Analysis of VTOL MAV use during rescue and recovery operations following Hurricane Katrina
Physical Description:
Book
Language:
English
Creator:
Pratt, Kevin S
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
UAVs
Semi-autonomy
Obstacle avoidance
Urban flight operations
Operator vehicle ratio
Dissertations, Academic -- Computer Science and Engineering -- Masters -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: There can be little doubt that Hurricane Katrina will always be remembered for the damage and devastation it caused. But it also provided the first opportunity for MAVs to be used and evaluated during Search and Rescue (SAR) as well as recovery operations. Researchers from The Center for Robot-Assisted Search And Rescue (CRASAR) made two separate deployments to areas affected by Hurricane Katrina: one during initial SAR operations and a second deployment during recovery operations 90 days later. Using data and observations from both of these deployments, this work draws four key findings about semi-autonomous Miniature UAV (MAV) operations in urban environments. These findings are intended to guide future MAV research as well as serve as a roadmap for the evolution from semi-autonomous to fully autonomous MAV capabilities. These findings are as follows: the minimum useful standoff distance from inspected structures is 2-5 m, omni-directional sensor capabilities are needed for obstacle avoidance, GPS waypoint navigation is unnecessary, and that these operations currently require three operators for one MAV.
Thesis:
Thesis ( M.S.C.S.)--University of South Florida, 2007.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Kevin S. Pratt.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 51 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001960585
oclc - 257601436
usfldc doi - E14-SFE0002209
usfldc handle - e14.2209
System ID:
SFS0026527:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001960585
003 fts
005 20080926142920.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 080926s2007 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002209
035
(OCoLC)257601436
040
FHM
c FHM
049
FHMM
090
QA76 (ONLINE)
1 100
Pratt, Kevin S.
0 245
Analysis of VTOL MAV use during rescue and recovery operations following Hurricane Katrina
h [electronic resource] /
by Kevin S. Pratt.
260
[Tampa, Fla.] :
b University of South Florida,
2007.
3 520
ABSTRACT: There can be little doubt that Hurricane Katrina will always be remembered for the damage and devastation it caused. But it also provided the first opportunity for MAVs to be used and evaluated during Search and Rescue (SAR) as well as recovery operations. Researchers from The Center for Robot-Assisted Search And Rescue (CRASAR) made two separate deployments to areas affected by Hurricane Katrina: one during initial SAR operations and a second deployment during recovery operations 90 days later. Using data and observations from both of these deployments, this work draws four key findings about semi-autonomous Miniature UAV (MAV) operations in urban environments. These findings are intended to guide future MAV research as well as serve as a roadmap for the evolution from semi-autonomous to fully autonomous MAV capabilities. These findings are as follows: the minimum useful standoff distance from inspected structures is 2-5 m, omni-directional sensor capabilities are needed for obstacle avoidance, GPS waypoint navigation is unnecessary, and that these operations currently require three operators for one MAV.
502
Thesis ( M.S.C.S.)--University of South Florida, 2007.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
590
Adviser: Robin R. Murphy, Ph.D.
500
Title from PDF of title page.
Document formatted into pages; contains 51 pages.
653
UAVs.
Semi-autonomy.
Obstacle avoidance.
Urban flight operations.
Operator vehicle ratio.
690
Dissertations, Academic
z USF
x Computer Science and Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2209



PAGE 1

AnalysisofVTOLMAVUseDuringRescueandRecoveryOperationsFollowingHurricaneKatrinabyKevinS.PrattAthesissubmittedinpartialfulllmentoftherequirementsforthedegreeofMastersofScienceinComputerScienceDepartmentofComputerScienceandEngineeringCollegeofEngineeringUniversityofSouthFloridaMajorProfessor:RobinR.Murphy,Ph.D.NagarajanRanganathan,Ph.D.JennyBurkePh.D.DateofApproval:November8,2007Keywords:UAVs,semi-autonomy,obstacleavoidance,urbanightoperations,operatorvehicleratiocCopyright2007,KevinS.Pratt

PAGE 2

cCopyrightbyKevinS.Pratt,2007ThisworkislicensedundertheCreativeCommonsAttribution-Noncommercial3.0License.Toviewacopyofthislicense,visit:http://creativecommons.org/licenses/by-nc/3.0/orsendaletterto:CreativeCommons543HowardStreet,5thFloorSanFrancisco,California94105,USA

PAGE 3

Dedication Tomyparentsandmysistersfortheirardent,unfailing,andutterlygenuinesupport.Thisthesisisoneofthemanythingsthatwouldnothavebeenpossiblewithoutyou.

PAGE 4

Acknowledgments ThismaterialisbaseduponworksupportedbytheNationalScienceFoundationunderSGERCMS0554059.Anyopinions,ndings,andconclusionsorrecommenda-tionsexpressedinthismaterialarethoseoftheauthorsanddonotnecessarilyreecttheviewsoftheNationalScienceFoundation.ThisdocumentwastypesetusingLATEX2"andissetin12ptTimesNewRoman.FormattingisaccomplishedwithUsfmanus.cls,aLATEXclassforcreatingUSFmaster'sthesesandPhDdisertations.TheoriginalclasswaswrittenbyChrisSperandiandHiman-shuGohelandhasbeenadaptedbyAaronGage,JenCarlson,andMattLong.Theorig-inalversionisavailablehere:http://www.eng.usf.edu/sperandi/LaTeX/index.shtmlAndalsoaspecialthankstotheentirecrewoftheHancockCounty,MississippiEOC.Eventhoughtheythemselveswerestilllivinginamuddyeldthreemonthsafterthehurricanehit,everyonetherewasenormouslygraciousinprovidingeverythinginclu-dinghotfood,internetaccess,aplacetoparkourRV,andeverythinginbetween.

PAGE 5

TableofContents ListofTables iii ListofFigures iv Abstract vi Chapter1Introduction 1 1.1HurricaneKatrina 2 1.2ResearchQuestions 3 1.3Contribution 3 1.4ThesisOrganization 4 Chapter2RelatedWork 5 2.1SARRobotics 5 2.2VTOLUAVUrbanOperations 6 Chapter3InitialTaskAnalysis 9 3.1WorkDomains 9 3.2RequiredOutput 10 3.3AirspaceOpportunities 10 3.4AirspaceProblems 11 Chapter4Approach 12 Chapter5EquipmentandTeamOrganization 14 Chapter6Flights 17 6.1HancockCountyEOC 17 6.2CasinoMagic 19 6.3GulfportGrandCasino 21 6.4HardRockCasino 22 6.5IsleofCapri 24 6.6PresidentCasinoBarge 27 6.71550BeachBlvd. 31 i

PAGE 6

Chapter7Results 34 7.1VehicleStandoff 34 7.2ObstacleAvoidance 35 7.3GPSWaypointNavigation 36 7.4Operator:VehicleRatio 38 7.5AdditionalFindings 38 7.6ProposedVehicleFeatureList 41 Chapter8Conclusion 43 8.1Summary 43 8.2FutureWork 44 References 47 ii

PAGE 7

ListofTables Table1.Locationanddatesforsurveymissions. 13 Table2.Location,date,duration,estimatedaveragealtitude,andestimatedaver-agestandoffdistanceofallights. 18 Table3.FeaturelistforUSARMAVreferencedesign. 42 iii

PAGE 8

ListofFigures Figure1.Verticalproleofanurbanstructuralinspectiontaskoverlayedwithas-setoperationzones. 11 Figure2.MapofBiloxi,MSshowinglocationsofstructuressurveyedbythe CRASAR teambetweenNovember26th-December5th,2005.MapcGoogle,2006. 12 Figure3.iSensysIP3 MAV plusPilotandMissionSpecialistoperationalequip-ment.ImageciSensys,Inc. 15 Figure4.Flightteamwithequipmentnotedandwithteammemberresponsibili-ties. 16 Figure5.Pictureofthe EOC inHancockCounty,MS. 19 Figure6.NortheastcorneroftheparkinggarageadjoiningCasinoMagic. 20 Figure7.PictureofthebargepilingsatCasinoMagic,andthedamagetothehotelcausedbythefree-oatingbargeoncebrokenloosebythestormsurge. 21 Figure8.DetailofpartiallyrepaireddamagetothesouthfacingsideoftheGrandCasino,whichabsorbedthebruntofthewindandstormsurgedamage. 22 Figure9.OverheadpictometryoftheGulfportGrandCasino. 23 Figure10.Pictureoftheeasternhalfofthetwostackedbargesleaningagainsttheland-sidebuildingattheHardRockCasino. 25 Figure11.Detailedpictureofdamagetoaguestroombalcony. 26 Figure12.CollapsedportionoftheIsleofCapriparkinggaragedamagedbyaloosebarge. 27 Figure13.YachtClubdockusedtolaunchthe IP3 tosurveydamageattheIsleofCapri. 28 Figure14.I-90bridgeheadingeastfromBiloxi. 29 iv

PAGE 9

Figure15.ThePresidentCasinobargewhereitcametorestapproximately1kmdownthebeachfromitsoriginalmooring. 30 Figure16.ImagescomparingnominalandaberrantpilotcamimagesfromPresi-dentCasinobargesurvey. 30 Figure17.SouthfaceofthePresidentCasinobargeshowingexposedandman-gledmetalsuperstructure,steelbeamconstruction,andhangingwiresprimarilyonrstoorwhichcausedahighlevelofradiointerferencewhileyingnearthebarge. 31 Figure18.Pictometryof1550BeachBlvd. 32 Figure19.Eastfaceof1550BeachBlvd.showingcollapsedsecondooranddam-agetoconcrete/woodframejoint. 33 Figure20.TheIP3yingintoabuildingtoimageconcealeddamage. 36 Figure21.Anexampleofwhyin-ightredirectsweresocommon. 37 Figure22.Thethreedifferentviewsoftheight-teammembers. 39 Figure23.DevelopmentpathfortheimplementationofautonomyinVTOLMAVsforurbaninspectionapplications. 45 v

PAGE 10

AnalysisofVTOLMAVUseDuringRescueandRecoveryOperationsFollowingHurricaneKatrinaKevinS.PrattABSTRACT TherecanbelittledoubtthatHurricaneKatrinawillalwaysberememberedforthedam-ageanddevastationitcaused.Butitalsoprovidedtherstopportunityfor MAV stobeusedandevaluatedduringSearchandRescue SAR aswellasrecoveryoperations.ResearchersfromTheCenterforRobot-AssistedSearchAndRescue CRASAR madetwoseparatedeploymentstoareasaffectedbyHurricaneKatrina:oneduringinitial SAR operationsandaseconddeploymentduringrecoveryoperations90dayslater.Usingdataandobservationsfrombothofthesedeployments,thisworkdrawsfourkeyndingsaboutsemi-autonomousMiniatureUAV MAV operationsinurbanenvironments.Thesend-ingsareintendedtoguidefuture MAV researchaswellasserveasaroadmapfortheevo-lutionfromsemi-autonomoustofullyautonomous MAV capabilities.Thesendingsareasfollows:theminimumusefulstandoffdistancefrominspectedstructuresis2-5m,omni-directionalsensorcapabilitiesareneededforobstacleavoidance,GPSwaypointnavigationisunnecessary,andthattheseoperationscurrentlyrequirethreeoperatorsforone MAV vi

PAGE 11

Chapter1IntroductionOnAugust29th,2005HurricaneKatrinaravagedthecoastalregionsofLouisianaandMississippi.Thoughthenumberswouldnotbenalizedformanymonths,itwasthedeadliestandcostliestAtlantichurricaneinnearly80years[ 20 ].Inthedaysandmonthsfollowing,hundredsoflocal,state,andfederalorganizationsandthousandsofpeopleweremobilizedtorespondtothedisaster.OneoftheseorganizationswasTheCenterforRobot-AssistedSearchAndRescue CRASAR fromtheUniversityofSouthFlorida.Oneteamfrom CRASAR wasonstandbywhenthestormhit,andwasdeployedtoBaySt.LouisandPearlington,MStohelpwithemergencyresponsefromAugust30ththroughSeptember1st.Ninetydayslaterasecond CRASAR teamdeployedtoBiloxiandGulfport,MStoconductstructuralsurveyoperationsinsupportofongoingrecoverywork.Thisrstdeploymentbythe CRASAR teamwastherstuseofMiniatureUAVs MAV sduringaliveemergencyresponseandtheseoperationscombinedconstitutethebulkofknown MAV operationsforemergencyresponse,andpracticallyalloftheightsconductedtodatewithscienticresearchinmind.Asthesearesomeoftheearliestoperationsoftheirkind,theseightscontainmanylayersofdataregarding MAV ights.Oneofthemostimportant,however,iswhatcanbelearnedabout MAV ightinclutteredurbanenvironments.UnmannedAerialVehicles UAV shavebeenaroundforatleast20years,butthusfartheyhavebeenprimarilylargexed-wingvehiclesownataltitudeconductingIntelligence,Surveillance,TargetAcqui-sition,andReconnaissance ISTAR operations.Theuseof MAV s,particularlyrotary1

PAGE 12

wingVerticalTakeOffandLanding VTOL aircraft,atlowaltitudeinclutteredurbanenvironmentsisanewstyleofoperationswithverylittlepublishedliteraturediscussingmethodologiesandrequirementsforsuchoperations.1.1HurricaneKatrinaHurricaneKatrinamadeitsrstlandfallattheMiami-Dade/BrowardCountyborderinsouthernFloridaat2230UTConAugust25th,butitwasnotuntilitssecondandthirdlandfallsinLouisianaandMississippiat1110and1445UTConAugust29ththatitwreakeditsdevastation.WhiletransitingacrosstheGulf,Katrinageneratedsustainedwindsof280kmhmphwithaminimumcentralpressureof902mbar,makingitaCategory5hurricaneontheSafr-Simpsonhurricanescale,the6thstrongestAtlantichurricaneonrecord,and3rdstrongesttomakelandfallintheUnitedStates.BythetimeitmadenallandfallontheGulfCoastonthe29thithadreducedinintensitytoacentralpressureof928mbarwithwindsat194kmhmph,makingitamid-gradeCategory3hurricane[ 20 ].Butitwasnothighwindswhichledtothemassivedevastation;ratheritwastheex-tensiveandhigh-intensitystormsurgeandthesecondaryoodingitcaused.AsthestormmovedoutoftheEvergladesandintotheGulfonthe26th,itrapidlybecameaverybroadstorm,acharacteristicitmaintainedwellintotheMississippivalley.Atitsnallandfall,thestorm'smaximumwindsextendedanestimated55kmnmfromtheeye,whilehurricaneforcewindsextendedaminimumof139kmnmoutward.ThisbroadsizeanditsstrongintensityovertheGulfgeneratedastormsurgethatwasbothbroadaswellascomparitivelydeep,inadditiontosignicantwavesbuiltontopofthissurge.Themax-imumstormsurgemeasuredwas8.74m.8ftatPassChristian,MS,with6mftthroughmostofHancockCounty,MS,andupto5.8mftinNewOrleansitself.Ontopofthisstormsurgewasthegeneratedwaveaction,forwhichaNationalDataBuoyCenter NDBC buoysouthofDauphinIslandmeasuredamaximumwaveheightof16.8 2

PAGE 13

mftaboveMSL;thisisalsothehigheststormwaveeverreportedbya NDBC buoy[ 20 ].Itwasprimarilythissurgeandwaveactionwhichooded80%ofNewOrleans,over-toppedandlaterbreachedlevees,attenedneighborhoods,turnedstreetsintolumberyards,causedover$80billionUSDindamages,andledtothedeathsofover1,800people.1.2ResearchQuestionsThuswearepresentedwith MAV sseeingtheirrstoperationaluseinemergencyresponseandrecoveryoperationsinaverywideareadisaster.Fromthissituation,thisworkendeav-ourstoanswertwocentralresearchquestions.Whattechnicalcapabilitiesarerequiredtousea MAV platforminthisenvironment?Howdoesa VTOL MAV handleurbanightconditionsvs.freespaceight?Howareaerodynamics,control,telemetry,andothersystemsaffectedbyightinandaroundbuildings?Arethereanyfeaturesthatwouldmaketheseoperationseasieronthepilotandightteam?Arethereanyofthecanonical MAV featureswhichdonotperformwell,orarenotneededinthisenvironment?Whatisthemosteffectivepathtoevolve MAV sfromsemi-autonomoustofullyautonomous?Howshouldfeaturesberankedintermsofdifcultyandnecessity?Isaninspectiontypetaskevenamenabletoafullyautonomousplatform?Ifnot,whatactivitiesshouldbeleftashuman-in-the-looptasksandwhichcouldbeautomatedtomaketheoperationseasierandmoreeffective?1.3ContributionThisworkmakestwokeycontributionstotheresearchcommunity.Therstoftheseisahistoricalrecordoftheightsandoperationstowhichfutureresearcherscanrefer.By 3

PAGE 14

documentingeachight,thisthesiscanserveasadenitiverecordontherstuseof MAV sinanurbanemergencyresponsesituation.Thesecond,andmoreimportantcontributionhoweveristotheeldroboticscommunity.Thisdocumentdescribesthefeaturesneededforsemi-autonomous MAV operationsinclutteredurbanenvironments,andbydoingsocanhelpguidefutureresearchwith VTOL MAV s.1.4ThesisOrganizationTheremainderofthisdocumentisorganizedasfollows.Chapter 2 discussespreviousre-searchon MAV ightoperationsinurbanenvironmentsinthecontextofdisasterresponseoperations.Chapter 3 presentsaninitialtaskanalysiscoveringthedifferenttypesofemer-gencyresponsescenariosandthedifferentimplicationsandrequirementsfor MAV swhileChapter 4 discussesthemethodologiesemployedincollectingandanalysingthedata.Fol-lowingthat,Chapter 5 describestheequipmentemployedaswellashowtheightteamwasorganizedwhileperformingtheoperationsandChapter 6 providesarecordofeachoftheindividualightsthemselves.Finally,Chapter 7 synthesizesthisinformationtodrawseveralconclusionsabouttheneededfeaturesandthepathforwardinevolvingurban MAV sfromsemi-autonomoustofullyautonomousplatforms. 4

PAGE 15

Chapter2RelatedWorkToproperlycontextualizethiswork,itisnecessarytoreviewworkfromtwoprimarysubjectareas;SearchandRescue SAR roboticsand VTOL MAV autonomyandurbanightoperations.2.1SARRoboticsTheuseofrobotsofanydescriptionduring SAR operationsisstillarelativelyneweld,butthereissomeprecedentwiththeuseofUnmannedGroundVehicles UGV sduringeldexercisesaswellasmasscasualtyincidentresponses.Therstuseof UGV sduringanactualresponsewasattheWorldTradeCenter WTC terroristattackinNewYorkCity.Micirediscussesthedifferentrobotdropsatthedisastersite,howandwheretherobotswereusedandnotused,aswellashowtherobotsfailedandwhatneedstobedonetocorrectthesefailures[ 24 ].HeproposessevenkeyareasthatshouldbeaddressedforfutureUrbanSearchAndRescue USAR UGV work,namely:stovepipesystemdevelopmentasexempliedbyincompatibleimageprocessingsystems,lackofautomatedtethermanage-ment,weakwirelesscommunicationssystems,self-congurationcapabilitiesforpolymor-phicrobots,SimultaneousLocalizationandMapping SLAM capabilitiesforthe UGV s,andnally,assistednavigationforthe UGV stofacilitateoperatormobilitywheninsidethepile.Blackburn,Everett,andLairdalsodiscusstheroboticresponsetothe WTC disasterbytheNavy'sSpaceandNavalWarfare SPAWAR SystemsCommandinSanDiego[ 2 ].Boththe SPAWAR teamandtheUSFteamwhereMicirewasamemberwerecoordi5

PAGE 16

natedunderthedirectionof CRASAR whileat WTC .Whileallsiteteamswereplacedunder CRASAR directionwhileonsitetheyoperatedessentiallyindependentlyandsawdifferentsituationswhenoperating.The SPAWAR reportidentiescognitivetunneling,situationalawareness,mobilitylimitationsofthe UGV s,andhighcognitiveloadfortheoperatorpartiallyinducedbylackofsemi-autonomoussupportcapabilitiesonthe UGV saslimitingfactorsduringthedeployment[ 2 ].Afterthisrstrobotic SAR deployment,robotteamshaveparticipatedinseveraldisas-terresponseexercisestofurtherunderstandhowtheycanbestbeusedinthesesituations.Gageetal.describetheexperiencesofarobotteamwhichparticipatedintheShadowBowl2003exerciseinSanDiego.Thereissomediscussionoftheactualon-siterobotdrops,buttheprimaryfocusofthisworkwasintestingoutroboticreachback 1 methodologiesandstrategies[ 9 ].Assemblingobservationsfromseveraloftheseresponses,MurphygivesathoroughdescriptionofcurrentFEMATaskForce TF commandstructureanddisasterresponseprotocolandhow UGV swouldneedtobeintegratedintobothofthese[ 25 ].Shegoesontoproposeadomaintheoryforrobotoperationandthestructureforinformationowthroughthe TF andIncidentCommandhierarchy[ 25 ].2.2VTOLUAVUrbanOperationsThethirdsubjectarea,urban MAV ightoperationsand MAV autonomydevolopment,hasseveralnotablesourcesaswell.Oneimportantworkdealingwithurban MAV operationsistheBlackhawkprojectdescribedbyGreenandOh[ 12 ].Thisprojectfocusedonusingahighly-maneuverablexed-wing MAV whichcanoperateasaxed-wingtoquicklytran-sitionacrosslongdistances,butcanalsogointoanautonomousprophangandoperateasa 1Gagedenesreachbackasfollows,`Reachbackreferstoestablishingcommunicationbetweentherst-respondersatthesceneofthedisasterandotherexpertsthatmaybegeographicallydistant[ 9 ].' 6

PAGE 17

rotary-wingvehicleforinspectiontasks.Onelimitationofthisworkisthattosuccessfullycompletetheprophangmaneuveritmusthaveathrust-to-weightratioofgreaterthan1T>WandT=W>1,whichfurtherlimitsthealreadystringentpayloadrestrictionsontheaircraft.Todeterminetheusefulnessofsuchavehiclefora USAR taskitisimportanttodecideifthegainsinloitertimeandtransitionspeedoutweighthepayloadandstabilitylimitationsofsuchahybrid.Movingtotraditional VTOL UAV s,Shimetal.demonstratedsuccessfulautonomous VTOL navigationbetweensimulatedurbanobstacles[ 39 ].Thisworkusedmultiplescan-ninglasersattachedtoaYamahaRMAXhelicoptertodetectandavoidtheobstaclesatthehelicopter'saltitude.Whileanimportantstepforvehicleautonomy,thisworkhaslim-itedapplicabilitytothestructuralinspection, USAR ,andrelated MAV ightdomainsastheRMAXisfarfromman-packableandcurrentman-packableplatformsdonothavethepayloadcapacitytomountallthehardwareusedforthisdemonstration.Athirdprojectdealingwith MAV ight-opsinclutteredurbanenvironmentsistheAVATARprojectfromUSC[ 16 ].Thisworkcombinesstereoimagingtechniquesandoptic-owtonavigatearotary-wing MAV downthecenterofanurbancanyon.Thisraisesafewimportantquestions,namely:Howprevalentareurbancanyonsintheoperationalspace?andIsthecenterofthesecanyonswherewewanttobe,ordoesthevehicleneedtobeclosertoonebuildingoranother?ThemostpromisingworkinurbanoperationsandobstacleavoidanceisapaperbySchereretal.fromtheRoboticsInstituteatCMU[ 36 ].Usingascanningladarandanartful3-DdodgercomprisedofcompetingverticalandhorizontaldodgerbehaviorstheirRMAXsuccessfullycompleted1000runsatspeedsupto10m/sagainstalltypesofob-staclesfromtreesandbuildingsto6mmwires.Whiletheladarandthedodgerbehaviordoprovideanelegantsolutiontotheproblem,therearetwolimitationstothiswork.First,theladaristoolargeforaman-packableplatformitislargerbybothvolumeandweight 7

PAGE 18

thantheentireIP3platform.Secondly,theladarsystemonlyprovidesobstacledetectioninaforwardlooking60by40cone;excellentforforwardight,butinsufcientforfullthree-dimensionalcoverage.Thefourth,butalltogethermissing,subjectareaistheuseof MAV sduringdisasterresponseand USAR scenarios.TheauthorpresentedapreliminaryversionofthisworkduringAUVSI2006,buttherearenoothersourcesdescribingtheuseof MAV sinadisaster[ 30 ].AstheKatrinaightsweretherstuseof MAV sforemergencyresponse,thisistherstseriesofworksdescribingthistypeofuse. 8

PAGE 19

Chapter3InitialTaskAnalysisToeffectivelyconductstructuralsurveymissionsand,moreimportantly,beabletoanalyzetheplatformrequirementsforsuchatask,itisimportanttohaveaninitialun-derstandingofthetask.Understandingstructuralsurveyrequiresunderstandingtheworkdomaingoverningthenatureoftheoperationaswellasthekeytaskstobeperformed.3.1WorkDomainsWithintheeldofemergencymanagementtherearefourstandardphasesinadisaster:thetwopre-eventphases,mitigationandpreparedness,andthetwopost-eventphases,responseandrecovery[ 27 ].Thesetwopost-disasterphasesdeliniatetheworkdomainswhere MAV swouldbeusedforstructuralsurveyduringemergencymanagement.Duringtherescuephaseofaresponsethestructuralinspectionworkwouldbecloselytiedtospecicgroupsofon-scenerespondersman-packable MAV swouldlikelybeor-ganictoresponderteamsdirectlyprovidingstructuralviewstoteammembersforanalysisandevaluation.Theresponsephasebeginsimmediatelyafteranydisasterwiththebulkconductedinthehoursanddaysimmediatelyfollowing,andistypicallyconcludedwithinoneweekofanincident.Therecoveryphasebeginsoncetheresponsephaseisconcludedandcanlastmonthsoryearsdependingontheseverityoftheincident.Duringtherecoveryphasestructuralinspectionworkby MAV swouldmostlikelybecoordinatedbyinsuranceclaimsadjustersorbuildingrepaircontractors.Duringthelesstimecriticalrecoveryphaseoperationsthe 9

PAGE 20

datafromthe MAV wouldlikelynotbeuseddirectlyintheeldbuttransmittedtoremoteexpertsinareachbackscenarioevenwhenthisisdesirableduringtheresponsephasetherealityoflimitednetworkavailabilityandreliabilitymakethisdifcult.3.2RequiredOutputIneitherworkdomain,structuralinspectionwillneedtoprovidethefollowingtypesofdata.Bothplanviewandelevationviewswillneedtobeprovidedandlabeledconsistentlywiththetheterminologyemployedbytheexpertsviewingthedatarespondersorstructuralexperts.Forbothsetsofuserswideshotsthathelpestablishoverallsituationalawarenessaswellasdetailedshotsofspecicdamagedportionsareneeded.Alloftheseshotsmustbehigh-resolutionstillstoprovideenoughinformationtostructuralexpertsthisinitialhypothesiswaslaterdirectlyconrmedbythestructuralexpertsduringreachback.Videocanbeuseful,butonlyasawaytoestablishtheoverallscene,nottoevaluateindividualstructuralelements.3.3AirspaceOpportunitiesInthepast,structuralinspectiontaskshavetypicallybeenaccomplishedbygroundassetsormannedaircraft.Whileeachofthesemethodshavetheiradvantages,theyalsohaveanoverlappingsetoflimitations.AsshowninFigure 1 ,thismissingsegmentisthespacebe-lowFAAregulatedairspace,butabovewhatcanbeachievedwithground-basedresources.Thisspaceisperfectlysuitedfor MAV operations.Inparticular,rotary-wing MAV soper-atinginthisspacecanprovideallofthedatatypespresentedintheinitialtaskanalysis. 10

PAGE 21

3.4AirspaceProblemsFor UAV stoyinthiszonebetweengroundassetsandmannedaircraft,orindeedany-whereintheUSNationalAirspaceSystem NAS ,willrequireregulatoryactionbytheFederalAviationAdministration FAA .AccordingtoAC91-57andtheFebruary6th,2007claricationto91-57UnmannedAircraftOperationsintheNationalAirspaceSys-tembyNicholasSabatini, UAV sarenotpermittedtooperateinthe NAS forcommercialorresearchpurposes,exceptwhenauthorizedbyaCerticateofAuthorization COA ;andthusfar COA shaveonlybeenissuedtoDoDandDHS[ 41 35 ].Notingthatthisissuemostcertainlyexists,discussionandpotentialsolutionstotheissueareleftforotherwritings. Figure1:Verticalproleofanurbanstructuralinspectiontaskoverlayedwithassetoper-ationzones.Rotary-wing MAV sprovideincreasedcapabilitiesoverground-basedassets,andprovidethesecapabilitiesatalowercostandwithashortersensor-analystpaththanmannedrotary-wingaircraft. 11

PAGE 22

Chapter4ApproachFromNovember26th-December5th,2005ateamofresearchersfrom CRASAR de-ployedtoHancockCounty,MStoconductstructuralsurveysofmulti-storycommercialstructuresinBiloxiandGulfport,MS.Duringthistimetheteamconducted12differentmissionsat7differentsiteswithatotalof32ightsduringallmissions.Figure 2 showsthelocationsoftheseightsandTable 1 providesthedatesandlocationsofthedifferentmissions. Figure2:MapofBiloxi,MSshowinglocationsofstructuressurveyedbythe CRASAR teambetweenNovember26th-December5th,2005.MapcGoogle,2006. Duringeachmissionwhichincludedallconsecutiveightsatonelocationseveralpiecesofdatawerecollected.Theseincludedpre-ightandpost-ightmeteorologicaldatasets,ightteamvoicerecordings,ightteamdebriefs,videostreamsfrom4videocam12

PAGE 23

Table1:Locationanddatesforsurveymissions. FlightLocation DatesFlown HancockCountyEOC 11/30/200512/3/2005 CasinoMagic 11/29/200512/2/2005 GrandCasinoGulfport 11/30/2005 HardRockCasino 11/30/200512/2/2005 IsleofCapri 12/1/2005 PresidentCasinoBarge 12/1/200512/3/2005 1550BeachBlvd. 12/2/200512/3/2005 eras,andstillpicturesfromthepayloadcamera.Thisdata,particularlythedebriengsandtheightvideo,wasthenreviewedandanalyzedtoderivethendingsregardingtheop-erationalandtechnicalrequirementsfor MAV operationsinclutteredurbanenvironments.Inthisanalysistheteamdebriengswereusedastheinitialresultsandthevideoandotherdatasetswerethenusedtocorroborateoramendtheteammembers'commentsaboutightconditionsandvehicleactions. 13

PAGE 24

Chapter5EquipmentandTeamOrganizationThestructuralsurveymissionswhichformthefoundationofthisworkwereevalua-tionsofmultipleindependent,multi-storycommercialstructuresapproximately90daysafterHurricaneKatrinamadelandfallintheGulfCoastregion.Theprimaryfocusofthesemissionswasrecovery-phasesingle-structureverticalinspectionsandthusthemosteffec-tiveplatformforthisworkwasaman-packablerotary-wing MAV .ForthesemissionstheImagingPlatform3 IP3 MAV wasselected.The IP3 isacommerciallyavailableplatformproducedbyiSensys,Inc.,anInstituteforSafety,Secu-rity,andRescueTechnology ISSRT NSFIndustryCentermembercompany.The IP3 islooselybasedontheMikadoLogo-14andhasbeenmodiedforstability,runtime,andpayloadcontrol.Itisanelectricallypoweredhelicopterwhichhasa42V4200mAhS3PLithium-Polymerbatterypack,a1.35mrotorspan,1kgofpayloadcapacity,axedpilot-viewcameratoprovideincreasedsituationalawarenesstothepilot,a15minuteightendurance,a25cm3-axisinniterotationgimbal,andcanholdupto8separateimagingsystemsandupto62.4GHzwirelessvideotransmitters.Inadditiontothe IP3 ,boththepilotandthemissionspecialistwereoutttedwith72MHzwirelesscontrollers,2.4GHzwirelessvideoreceivers,Heads-UpDisplays HUD s,andvideocamerastorecordallreceiveddata.Figure 3 illustratesallofthisequipment.Tooperatethe IP3 systemathree-manightteamwasused.Theightteamconsistedoftheightdirector,whowasresponsibleforteamsafety,maintainingoverallsituationawareness,andforthemissionasawhole;thepilotwhowasresponsiblefortheaircraft; 14

PAGE 25

Figure3:iSensysIP3 MAV plusPilotandMissionSpecialistoperationalequipment.ImageciSensys,Inc. andthemissionspecialistwhowasresponsibleforthepayloadandgatheringthedatatargetedduringthemission.Eachteammemberwasresponsibleforadifferentlevelofsituationalawarenessandhadaninverselyproportionaldegreeofcomputermediation.Figure 4 showsthe IP3 equipmentinuseontheightteam,aswellastheteamorgani-zationandresponsibilitiesduringamission. 15

PAGE 26

aFlightTeamwithequipment bFlightTeamwithmissionresponsibili-tiesnotedFigure4:Flightteamwithequipmentnotedandwithteammemberresponsibilities. 16

PAGE 27

Chapter6FlightsWhileconductingstructuralsurveysduringHurricaneKatrinarecoveryoperationsinBiloxiandGulfport,MS,the CRASAR teamconducted32ightsat7sitesover8days.Whileinandofthemselvesthesearerelativelyinnocuousoperations,theycarrysomehis-toricalsignicanceasHurricaneKatrinaistherstdisastertoseetheuseofanytypeof UAV sduringeitherpost-disasterphase.Withthatinmind,thischapterwilldetaileachoftheseightlocations,themissionandsurveillanceobjectives,andtheconditionsencoun-teredateachsite.ThesedescriptionswillthenprovidethebasisfortheconclusionsdrawninChapter 7 .Tobeginthisdiscussion,Table 2 providesasummaryoftheightsandlistseachightandprovidestheightdurationeach.6.1HancockCountyEOCEventhoughitwasnotoneofthemulti-storycommercialsitesweweretheretophoto-graph,weewthemostnumberofightsattheHancockCountyEmergencyOperationsCenter EOC .Andwhileitwasnotaresearchtarget,theseightswerejustasimportant,astheygaveusanopportunitytoevaluateandtestourteamworkprocesses,opticsandvideosystems,andourcontrolmethods.Figure 5 showsanoverviewofthe EOC takenduringoneofourtestights. 17

PAGE 28

Table2:Location,date,duration,estimatedaveragealtitude,andestimatedaveragestand-offdistanceofallights.Averageighttimeis3:29andmedianighttimeis3:15.yDataunavailable. Date FlightLocation FlightNumber Durationm:sec EstimatedAverageAltitudem EstimatedAverageStandoffm 11/29/05 HancockEOC 1 y 15 y 2 y 25 y CasinoMagic 1 2:32 80 100 2 4:15 60 30 3 3:44 30 15 11/30/05 GrandCasino 1 2:28 120 40 2 2:42 100 70 3 0:34 20 25 HardRock 1 6:46 50 30 2 1:53 2 2 3 1:04 25 10 4 3:16 20 5 5 3:07 50 15 12/1/05 IsleofCapri 1 4:55 40 60 2 2:48 50 70 PresidentBarge 1 3:40 30 20 2 2:05 20 5 3 2:31 10 5 4 3:28 15 2 12/2/05 CasinoMagic 1 3:15 30 15 2 3:45 20 10 3 3:20 40 20 HardRock 1 9:32 30 15 2 2:27 40 20 1550BeachBlvd. 1 5:24 15 4 2 6:28 20 8 3 1:06 5 2 12/3/05 HancockEOC 1 y 30 y PresidentBarge 1 0:59 25 10 2 5:06 25 5 3 0:57 10 3 1550BeachBlvd. 1 7:00 15 4 18

PAGE 29

Figure5:Pictureofthe EOC inHancockCounty,MS.Thiswastakenduringourthirdtestightatthesite. 6.2CasinoMagicAsTable 1 notes,weconductedtwomissionsatCasinoMagic;therstoperationoncewewereinBiloxionNovember29thandagainonDecember2nd.Thetargetfortherstmis-sionwasapartiallydamagedparkinggarageattheedgeoftheCasinoMagicproperty.Theloweroorsofthegaragewerelledwithdebrisandassortedotsamthathadaccumulatedwhenthestormsurgehadrecededthroughthearea.Thismadeitdifculttoseethroughthestructureandtelliftherewasanydamagetotheinteriorverticalmembersoroorslabofthestructure.Usingthe MAV wewereimmediatelyabletogetaviewpointfromabovethestructureandevaluatethedeckstructure.Whilewewereyingduringtherecoveryphase,Figure 6 belowclearlyshowsthistypeofimagerycouldeasilybeusedtoquickly 19

PAGE 30

reconnoiterastructureandidentifyifanyonehadsoughtrefugeontheroofparticularlyusefulwhentheinteriorstabilityofthebuildingcannotbesafelyascertained. Figure6:NortheastcorneroftheparkinggarageadjoiningCasnioMagic.Partofasetofimagesweusedtoevaluatetheintegrityoftheupperdeck. DuringthesecondmissionatCasinoMagic,weturnedourattentiontothebargemoor-ingarea.Aswithseveralbargesinthearea,thestormsurgecausedtheCasinoMagicbargetobreakitsmooringsandoatfree.Oncefree,thewaveactionontopofthestormsurgeturnedthebargeintoabatteringramandcausedseveredamagetoadjacentstructures.InFigure 7 boththepilingsthebargewasoriginallymooredto,aswellaswhereitrammedthestructuralbeamsoftheadjoininghotelareclearlyvisible.ThepilingsshownhereareoneclearexampleoftheightpathobstaclesthatcanbeencounteredduringtheseoperationsseeSection 7.2 formoredetails. 20

PAGE 31

Figure7:PictureofthebargepilingsatCasinoMagic,andthedamagetothehotelcausedbythefree-oatingbargeoncebrokenloosebythestormsurge. 6.3GulfportGrandCasinoFollowingthisrstmissionatCasinoMagic,ournextmissionwasattheGulfportGrandCasino.Incontrasttothelong,lowparkinggaragebefore,theGrandCasinoisarelativelytallbuildingandonlyhadvisibledamageontheseawardsideadditionally,workcrewswerepresentonothersidesofthebuilding,soforsafetyconcernswelimitedourselvestothissideofthebuilding.AsnotedinSection 7.5 below,oneoftheimportantndingsfromtheseightswastheimportanceofthepilot-viewcamera.SinceGrandCasinowasthetalleststructure,andhencewasthesiteofthehighestights,itprovidedanexcellentop-portunitytotestthispilot-cam.Whenremotelypilotingsuchasmall MAV ,itbecomeseasytoloseperspectiveandorientationonthevehiclewhenyingatanyappreciabledistance.Flyingat100mnearlyoverhead,maintainingamentalmodelofthevehicle'sorientation 21

PAGE 32

becomesasignicantchallenge.Section 7.5 providesfurtherdetails,buttheseightspro-videdanexcellentopportunitytoutilizethispilot-camandtheperceptualaugmentationitcanprovide.Figure 8 belowshowsoneofthedetailshotsofdamagebeingrepairedonthesouth,sea-facingsideoftheGrandCasino,whileFigure 9 isoverheadpictometrysurveydataoftheentireGrandCasinostructure.Thispicture,andallthepictometry,waskindlyprovidedbytheHancockCounty EOC teamwhohadcollectedandorganizedallofthisdata. Figure8:DetailofpartiallyrepaireddamagetothesouthfacingsideoftheGrandCasino,whichabsorbedthebruntofthewindandstormsurgedamage. 6.4HardRockCasinoOurnextlocationwastheHardRockCasinoinBiloxi.Justasatseveraloftheothercasinosinthearea,thegamblingbargeshadcomelooseduringthestorm.UnlikethePresident, 22

PAGE 33

Figure9:OverheadpictometryoftheGulfportGrandCasino.Theblue-roofedbargewhichismissingfrominfrontoftheGrandcanbeseennearthetopoftheframe.Thiswasanunfortunatelycommonoccuranceupanddownthebeach.ImageiscHancockCounty EOC ,2005andisusedwithpermission. theMagic,andtheIsleofCapri,forexample,theHardRock'sbargeshadstayedconnedtotheirconcretemooringcorral.However,insteadofthetwobargesbeingmooredsidebysideastheyshouldbe,theyhadendedupcreatingatwo-layerlean-toagainstthelandsidehotelportionofthecomplex.Figure 10 showsonehalfofthetwobargesleaningagainstthebuilding.DuringtheightsattheHardRockCasino,oneoftheprimarysurveillance 23

PAGE 34

targetswastoinspectwhatwasaboveandbehindthesebarges,aswellasinspecttheinterfacebetweenthebargesandthebuilding,andhowstructurallystablethatwas.ThesecondinspectiontargetwhileyingattheHardRockwasthelowestexteriorguestroombalcony,whichhadbeenpartiallycollapsedbythemovementofthebarges.AsFigure 11 shows,thedamagewasnotparticularlysevere,butthistypeofdamagecollapsedconcreteslabwithattachedinternalrebarreinforcementprovidedanexcellentopportunitytotestthecapabilityofthe IP3 andtheonboardopticstotargetandprovidedetailedimageryoffocuseddamage.Forthesesurveysofthebarges,theonlyavailablelaunchinglocationwasfromthepierthathadonceencircledthebarges.Asthepieritselfwasonly2mwide,itnotonlyprovidedalimitedlandingarea,butlimitedmobilitytotheightteamwhileoperating.AswiththedocklaunchattheIsleofCapriandthedebrisencounteredaround1550BeachBlvd.,thislimitedlaunchlocationonlyservedtoreinforcethelessonslearnedregardingsiteaccessandrestriction.Section 7.5 containsfurtherinformationontheimplicationsoflimitedsiteaccess.Inadditiontothebargesandtheassociatedarea,whileattheHardRockwealsoewatasecondarylocation.AsFigure 20 shows,weusedthe IP3 topenetratethedamagedsectionofthehotelandinspectstructuralmembersfromunderneath.Asisevident,thisprovidedsomeverydifcultyingforthepilotandunderscoredtheneedforanomni-directionalobstacleavoidancesystemasdescribedinSection 7.2 .6.5IsleofCapriTheworstdamagebyfree-oatingbargebatteringramsoccuredattheIsleofCapricasino.AttheIsleofCaprithebargehadbeenmooredbetweenthehotelandtheparkinggarage.Oncethebargebrokeloosefromitspilingsthestormsurgepushediteastwardintotheparkinggarage,whereuponthewaveactionrepeatedlydrovethebargeintothestructure, 24

PAGE 35

Figure10:Pictureoftheeasternhalfofthetwostackedbargesleaningagainsttheland-sidebuildingattheHardRockCasino.Thewhitedeckontheleftistherstbargewithallsuperstructureremovedbythestorm,andthereddeckisthe2ndoorofthesecondbarge.Therustedsteelframemembersarethesouth-facingwallofthebuildingsection. ultimatelycausingthecollapseofapproximatelyonequarteroftheparkinggarage.Figure 12 showsthismulti-storyhangingcollapseontheparkinggarage.Inordertosurveythisdamagetotheparkinggaragewewererequiredtolaunchfromtheyachtclubdockbehindtheoriginalmooringposition.AsdetailedinSection 7.2 ,thepilingsonthisdockprovidedaveryclearcasefortheneedforobstacleavoidancetech-nologiesonasemi-autonomous MAV platform.Figure 13 belowshowsthedockwiththelaunchinglocation,aswellasthedock'sorientationtotheparkinggaragetobesurveyed.ThesurveyattheIsleofCaprialsogaveusanopportunitytotestanothermodalityoftheinspectioncapabilityofthe IP3 .Thoughwewereworkingduringtherecoveryphase,andthistechniquewouldbemoreusefulduringtheresponsephase,Figure 14 belowshows 25

PAGE 36

Figure11:Detailedpictureofdamagetoaguestroombalcony.Thesecondbargecanbeidentiedbytheredpatternedcarpetinthelowerleft.Thispictureisapproximately50mwestoftheimageshowninFigure 10 averyclearshotofthedamagetothedeckingoftheI-90bridgeeastboundoutofBiloxiFigure 12 showsthebridgeinthebackgroundbehindthegarage.Fromtheground,wehadnowaytodeterminewhatconditionthebridgewasin,andwereweoperatingasrespondersduringthehoursanddaysfollowingKatrinawewouldhaveneededtomakeourwaytothebridge,determinethatitwasimpassible,andbacktrackandcomeupwithanalternateroute.Eventhoughthebridgewasashort2kmawayfromthedock,duringtheKatrinaresponseprogressingthatfarsometimestooktheFEMA SAR teamshoursastheypickedtheirwaythroughthetrashanddebristherecedingstormsurgehadleftinmostofthestreetsandroads[ 40 ]. 26

PAGE 37

Figure12:CollapsedportionoftheIsleofCapriparkinggaragedamagedbyaloosebarge.TheI-90bridgeFigure 14 isvisibleontheleftsideoftheframe. 6.6PresidentCasinoBargeWhilemanyofthecasinobargesbrokelooseandlandedelsewhere,thePresidentCasinobargecoveredthemostdistanceduringthestorm,endingup1kmwestand100minlandofitsstartinglocation.WhatnallystoppedthebargefromdriftinganyfurtherwasitscollisionwiththeroofoftheMotel6.AsFigure 15 shows,oncethewaterrecededthebargewaslefthighanddrynexttothemotel.DuringourightsaroundthePresidentCasinobargeweconcentratedprimarilyonthethreeclosestsidesofthebargeandtheinterfacebetweenthebargeandthemotelseeningreaterdetailinFigure 21 .Eachofthethreesidesweewonthebargepresentedauniqueightconditionandgaveusanopportunitytolearnsomethingnew.Beginningthediscussionontheleftsideofthebarge,groundaccessfortheightteamisanimportantissue.Welaunchedthisight 27

PAGE 38

Figure13:Yachtclubdockusedtolaunchthe IP3 tosurveydamageattheIsleofCapri.Thelaunchhazardpresentedbythesupportpolesisclearlyevidentinthisshot. fromtheparkinglotatthebottomoftheframe,thusinordertogetagoodviewoftheentirelengthofthebargewehadtoyatasomewhatextendedrange,limitingthepilot'sperspectiveandperceptionofthehelicopter.Atonepoint,aradioglitchtothecyclicthrewthehelicopteroutofastablehoverandintoaleftroll.Asnotedbythepilotimmediatelyfollowing,hadwenothadthepilotcamatthatpoint,itwouldhavebeenimpossibletorecoverthehelicopter.Figure 16 showsacomparisonbetweenthenominalviewfromthepilotcamandtheimagepresentedduringtheradioglitch.Movingcounter-clockwisearoundthebarge,thelowersideofthebarge,whereitisrammedagainstthemotel,presentedaveryclearcaseforin-ightredirectionandmodi-cationtotheestablishedightplan.Fromthegrounditisobviouslyimpossibletotelltheextentofdamageatthisseam;themajorityoftheroofcouldbecavedinortherecould 28

PAGE 39

Figure14:I-90bridgeheadingeastfromBiloxi.Ascanbeseenacrosstheentirespan,thecombinedstormsurgeandwaveactionliftedandthencollapsedallofthedecksegmentsofthebridge.ThisimagewastakenfromtheoppositesideoftheCapriparkinggarageandillustratesavaluableusecasefor MAV sconductingtacticalreconnaissancefor SAR unitsduringtheresponsephase. benodamageatall,anduntilthehelicoptercanmakeaninitialsurveyofthis,itisnotpossiblefortheightteamtodeterminehowmuchtimeneedstobespentsurveyingthisdamage.Figure 21 inSection 7.3 showsadetailshotofthisjoint.Continuingaroundtotherightsideofthebarge,thisfacepresentedtheworstradiointerferenceofalltheightlocations.Throughallthemissionsduringthetrip,radioin-terferencewasencounteredatagreatlyelevatedratecomparedtonormaloperationsanaverageof1to2hitsperightwasencountered,whereasnormalightregisters1to2hitspermonthofdailyying,butthesteelbeamconstruction,twistedsuperstructure,andmassofwiresencounteredonthissouthfaceofthebargepresentedevengreaterthreatofreectedandmulti-pathinterference.Figure 17 showsthesouthfaceofthebargewithexposedmetalsuperstructureandotherinterferencehazards. 29

PAGE 40

Figure15:ThePresidentCasinobargewhereitcametorestapproximately1kmdownthebeachfromitsoriginalmooring.Thebargeisthestructureontherightoftheframetoppedbytheredtowers.TheremainingbuildingsarepartofMotel6.ImageiscHancockCounty EOC ,2005andisusedwithpermission. aNominalpilotcamview bPilotcamviewduringahard-bankednosedivecausedbyaradioglitch.Figure16:ImagescomparingnominalandaberrantpilotcamimagesfromPresidentCasinobargesurvey.Atthetimeoftheglitchthe IP3 wasattheedgeofthepilot'sper-ceptualrange.Asnotedbythepilotduringthemissiondebrief,withoutthepilotcamprovidingahorizonline,theglitchwouldhavecausedthelossofthecraft. 30

PAGE 41

Figure17:SouthfaceofthePresidentCasinobargeshowingexposedandmangledmetalsuperstructure,steelbeamconstruction,andhangingwiresprimarilyonrstoorwhichcausedahighlevelofradiointerferencewhileyingnearthebarge.Aswithmucheld-work,andparticularlyemergencyresponseoperations,conditionsareoftenverydifferentfromcontrolledtesting,andmostoftenfortheworse. 6.71550BeachBlvd.Thenallocationinspectedwas1550BeachBlvd.,atwo-building4-storyresidentialblock.Thiswastheonlyresidentialbuildinginspectedandassuch,theonlywood-frameasopposedtosteelframestructureweinspected.Thetwobuildingsatthesitewerestackedinaroughlynorth-southorientationwhichconsequentlycausedthesouthernbuildingtoactasabreakwaterandabsorbthemajorityofthewaveactiondirectedatthesite.Whenwearrived,allthatremainedofthesouthernsitewastheconcretefoundation,andthenorthernbuildingwaslistingapproximately15degreestothenorth,reectingthestormsurgeandwaveactionpressure.Figure 18 showstheoverheadpictometryof1550 31

PAGE 42

Beach,withtheconcretepadfromthedestroyedbuildingvisiblenearthetopofthepicture,thesecondbuildingvisibleinthemiddle,andtheaccumulatedotsamanddebrisvisiblethroughouttheimage,butparticularlyjustbelowthesecondbuilding.Thisdebrisaccumulation,aswellasthedamagetoallsidesofthebuilding,illustratedinFigure 19 ,madethislocationaprimedemonstrationofthenecessitytoconductoperationsasmultipleshortights,ratherthanonelongightpersite.Whileitispossibleforthepilottoperformlimitedrelocationswhileying,thelargequantitiesofdebrisencounteredbytheFEMAUSARteamsmadetransitioningaroundthebuildingimpossiblewhileinight[ 40 ].Theoptimumoperationaltempoevolvedtobetotakeoffandinspectoneface,land,relocate,andlaunchtoinspectthenewbuildingface. Figure18:Pictometryof1550BeachBlvd.Showntheconcretefoundationofthedestroyedbuildingnearthetop,theremainingbuildinginthemiddle,andtheaccumlatedotsamanddebrisjustbelowit.Imageislookingsouth,withthebeachvisiblealongtheverytopoftheframe.ImageiscHancockCounty EOC ,2005andisusedwithpermission. 32

PAGE 43

Figure19:Eastfaceof1550BeachBlvd.showingcollapsedsecondooranddamagetoconcrete/woodframejoint. 33

PAGE 44

Chapter7ResultsOriginallywesetouttwocentralresearchquestionstohelpstructurethediscussion,namely:Whattechnicalcapabilitiesarerequiredtousean MAV platforminthisenviron-ment?,andWhatisthemosteffectivepathtoevolve MAV sfromsemi-autonomoustofullyautonomous?Reectingonourightoperationsinlightofthesetwoquestions,wedrawfourcentralconclusionsaboutconductingurbaninspection-typetasksinclutteredurbanenvironments.Theseoperationsshowthatastandoffrangeof2-5mistheminimumstand-offdistancerequiredbytheseoperationsthereisnooperationalrequirementfor MAV stooperateanycloserthanthistothetargetstructure,thatomni-directionalobstacleavoid-anceisnecessarytomove MAV sfromteleoperatedtosemi-autonomouscapabilities,GPSwaypointnavigationisnotarequiredfeatureforstructuralinspectiontasks,andthattosafelyandeffectivelyconductinspectionmissionsathree-manightteamisrequiredforeach MAV .7.1VehicleStandoffInastructuralinspectiontasktheclarityanddetailoftheimagesproducedarecrucialtosuccessfulanalysisbystructuralexperts.Qualitativelythisrequiresthe MAV tobeascloseaspossibletothestructurebeinginspected.Giventhepoweroftoday'scommonlyavailableopticstherealityisthatthereisaneasilyachievableminimaforthisrequirement.Duringtheseightsthe IP3 wasoutttedwithasimpleconsumer-gradeCommercialOffTheShelf COTS 5Megapixeldigitalcamera.Evenwiththisentry-level COTS solution, 34

PAGE 45

theimagestakenfromthis2-5mdistancehadampledetailandclarityforstructuralexpertstoperformtheiranalysis.Bymovingtoanimprovedopticspackagethisdistancecouldbecorrespondinglyincreasedthe IP3 gimbalwasdesignedtosupportaCanonEOS5Dcamerabodyandatelephotolens.The5Dshoots12.8megapixelimages.7.2ObstacleAvoidanceForimagingpurposesitwasbesttoplacetheimagingtargetsinthelowerhalfoftheforwardquadrant;thisdoesnotmeanhowever,thatthatisfromwhereallightobstaclesapproachedthevehicle.DuringtheBiloxiightsthepilothadtobeawareofandavoidobstaclesencroachingontheaircraftfromallangles,justasanysemi-autonomous MAV wouldneedtobe.Itiseasytotakepicturesoftargetsthatare12o'clocklow,butaswithmanygenerationsofghterpilots,remembertowatchyoursix.Figure 20 showsthemostcomplexenvironmentencounteredwiththe IP3 .Inthisimagethe IP3 wasyingintoabuildingtoimageastructuralbeamthathadbeencompromisedbyrepeatedimpactsfromabargethatendeduprestingonthestructure.Abovethe IP3 isasolidsteelceilingaswellashangingwiresandceilingtilesupports,belowitarecautiontapeaswellasa2x4barrier,andencroachingfrommultiplesidesareatrashcompactorandseveralstructuralsteelbeams.Additionallythehangingceilingsupportsandthecautiontapewerelightenoughthattheyweremovinginthewindandtherotorwashofthe IP3 .Atotherlocationsobstaclessuchastrees,agpoles,electricandphonelines,buildingoverhangs,anddamagedbuildingsuperstructurewereallpresentintheightpath.Whilethevehiclewillnotneedtocomewithin2-5mofitsintendedtarget,anysemi-autonomous MAV willneedtobeabletosuccessfullydetectandavoidalltypesofobstaclesapproachingfromallangles. 35

PAGE 46

Figure20:TheIP3yingintoabuildingtoimageconcealeddamage. 7.3GPSWaypointNavigationInexistingautonomousUAVs,GPSisaverycommonnavigationsolution.Asresearchersseektodevelopsemi-autonomous MAV s,GPSisaclearchoiceforinclusioninthesesys-tems,butbeforeextensiveeffortisinvesteditisimportanttoconsiderifthisisanappro-priatestep.For MAV susedinstructuralsurveytasks,GPSwaypointnavigationwouldnotbeacommonlyusedfeatureandshouldnotbeacentraldevelopmenttaskforsuchsystems.Inthisworkdomaininspectiontasksareverymuchhuman-in-the-looptasks;notonlytomaintainoperator,bystander,andvehiclesafetybutalsotoevaluatetheresultsinreal-time.Asitisbynecessityahuman-in-the-looptaskandoperatorsarealreadyevalu-atingthereturningdatastreams,theywillinevitablyseenewthingswhichwereoccludedfromtheirground-basedpreightpositions,modifytheordertaskswillbeaddressed,or 36

PAGE 47

evensimplyrequirethe MAV tostayonagiventaskforlongerthananticipated.InshorttheBiloxiightsshowedthatstructuralsurveyworkisaverydynamictaskandthatightplansregularlychangedassoonasthe IP3 tookoff.Tothedegreetheightplanchangesonceairborne,navigatingtoaxedsetofGPScoordinatesbecomesrapidlynotuseful.Figure 21 showsacasewhereitwasnotpossibletodeterminethenatureofthedamagefromthegroundduringthepre-ight,andithadtobeevaluatedfromthehelicopteroncetheighthadbegun.Thedamagecouldhavebeensupercialornon-existent,notrequir-ingexaminationwiththe IP3 ,oritcouldhavebeenverysevereandrequireimmediateandthoroughevaluation. Figure21:Anexampleofwhyin-ightredirectsweresocommon.Thenatureandextentofthedamagecouldnotbeevaluatedbeforehandandhadtobeevaluatedin-ighttodeterminewhatrequiredfurtherdocumentationandanalysis. InUAVsystems,particularlyxed-wingvariants,GPSwaypointnavigationisacriticaltool,andiftheightplanchanges,thewaypointscanbemodiedonthegroundstationand 37

PAGE 48

thenewsetuploaded.Thispresentstwoproblemswhenappliedto MAV sinastructuralinspection-typetask.Therstisthatouraverageighttimewasbetween10and15min-utes,soanyupdateprocessthattakes1minute,oreven30secondsconsumesasignicantportionoftheighttimewithunnecessaryoverhead.Thesecondproblemisthatchangesofthisnaturewouldrequiresomeformofcontrolstationcomputer,andforasystemtobeusableitmustbeentirelyman-packable;operatorsinthisworkdomainaremobileduringights,notseatedatacomputerterminals.7.4Operator:VehicleRatioItisbothnormalandnaturaltowanttomaketheoperator:vehicleratioaslowaspossible.Forbothsafetyandeffectivenessreasonsworkingwithasemi-autonomous MAV tocon-ducturbanightoperationsrequiresaminimumofthreeoperatorstoonevehicle.Whileconductingight-opsinBiloxi,athree-manightteamwasused:Pilot,MissionSpecial-ist,andFlightDirector.AsFigure 22 illustrates,theprimaryreasontheserolescannotbecombinedwithoutaseveredegradationofperformanceisthelackofinformationalover-lapbetweenthethreepositions.Whileallteammembersarelookingatthesamescene,eachmemberseessomethingdifferent.Inrelationtotherobot,theMissionSpecialisthasapurelyegocentricviewpoint,theFlightDirectorhasastrictlyexocentricview,andthePilotalternatesbetweenexocentricandegocentricthoughaltogetherdifferentegocentricandexocentricviewsthantheothercrewmembers.7.5AdditionalFindingsInadditiontothefourprimaryndings,thereareseveralotherimportantobservationstobedrawnfromtheseights.Forsafetyreasonsallightsmustbeconductedwithinlineofsight;notonlymusttheightdirectorbeawareofallpeopleintheightarea,butduring 38

PAGE 49

Figure22:Thethreedifferentviewsoftheight-teammembers.Eachteammemberhasadifferentfocusandadifferentdegreeofcomputermediation,makingrolecombinationdifcultandundesirable. anemergencythepilotmustbeabletosafelyguidethe MAV toalandingzone,andthustosurveymultiplesidesofabuildingrequiresmultipleshorterights,ratherthanasinglecombinedightforanentirebuilding.Aslongasthe MAV canberapidlyrefueledaswithbatterychangeforanelectric MAV ,thetotalsystemendurancebecomeslessofafactor.Anothernotablelessonwasthattobeaneffectiveeldteam,oneoftheteammembersmusthavesomedomainexpertisetohelpguideanddirecttheteamintheeld.Particularlyinarecovery-phasestructuralsurveytaskwithreachbacktoremotestructuralexperts,itwascrucialtohaveoneteammembertheightdirectorinthiscasewhohadformalstructuraltrainingandcouldserveasanintermediary/translatorbetweenthetwogroupsandasanon-siteexperttodirectthesurveymissions.Thefocusofthissetofinspectionswascommercialstructures,allbutoneofwhichweresteelframeconstructionwithvaryingdegreesofadditionalmetalinthesidingand 39

PAGE 50

roongmaterials.Unsurprisingly,usingconsumergradewirelesscommunicationsequip-mentinthissteeljungleledtoverynoticeableinterferenceandsignallossproblems.Anyprofessional-grade MAV systemdesignedtooperateinurbanenvironmentsmusttakethisintoaccount.Onelessonthatwasfortunatelyonlyaconrmationofourinitialdesignwasthesuccessofthepilotcamsystem.AsillustratedinFigure 16 above,thepilotcamsystemprovidedaverysimplebuteffectivewayofaugmentingthepilot'sactivestateinformationaboutthe MAV whileitwasintheair.AtboththeGulfportGrandCasinoandthePresidentCasinobargethepilotcamtemporarilybecametheprimaryightsensor,allowingthepilottomaintaincontrolofthecraftinsteadofrecoveringfromacrashlandingandthelossoftheaircraft.AttheGrandCasinotheheightofthetargetsurveyedrequiredthe IP3 operateattheedgeofthepilot'srange,puttingthepilotcamintoanactivesupportroletomaintainastablehover.AtthePresidentCasinobarge,the MAV encounteredasevereradioglitchtothecyclic,causingahard,nose-downroll.Inthiscasethepilotcamwastheonlysensorcapableofprovidingthepilotsufcientinformationtorecoverfromtheglitch.Anallessontoconsideristhatsiteaccesswasaveryimportantconsiderationduringthesestructuralinspectiontasks.Siteaccessincludesboth MAV landingzonesaswellaspersonnelpositioningduringtheights.Duetobothsafetyconcernsandthedifcultyofmovementbothnearcompromisedstructuresandthroughsuchawide-areadisaster,goodlandingzonesandteampositionswerebothdifculttocomeby,andthosethatwereavailablewereoftenfarawayand/orsuboptimal.Inshort,comparedtoatypicalhobbyRCighttheseightswerelaunchedandrecoveredfromsmaller,moreconnedlandingzonesandconductedatlongerstandoffranges.Toaddressthisproblem,solutionsthatcanextendthepilot'seffectiveoperatingrangeandprovideincreasedcontrolofthe MAV areneeded.Figure 18 inSection 6.7 aboveshowsthedebristhataccumulatedaround1550BeachBlvd. 40

PAGE 51

7.6ProposedVehicleFeatureListHavingnotedthesefourcentralconclusionsfromtheights,itispossibletostepbackandsynthesizetheseresultsandotherobservationsandproposeafeaturelistforfuturevehicles.Thuswehavethefollowingreferencedesignfora VTOL MAV foranexampleurbaninspectionapplication,suchasemergencyresponse USAR .Suchan MAV shouldbeelectricallypoweredandman-packableanditshouldhaveapilotcam,stabilizedhover,bothmanualandsemi-autonomousightmodes,videoandhigh-resolutionstillpayloadoutput,obstacleavoidance,upgradedradiolinks,payloadstabilization,payload`return-to-center',andahands-offhover/`hoverhere'capability.Table 3 listseachofthesefeaturesandtheoriginofthisrecommendation. 41

PAGE 52

Table3:FeaturelistforUSARMAVreferencedesign. MAVFeature Source Electric EaseoftransportabilityFieldrechargecapability Man-packable MobilityStandardFEMAequipmentsizingInterviewwithFEMAUSARspecialist[ 40 ] Pilotcam Pilotoutbrief11/30GrandCasinoFlight112/1PresidentBargeFlight2 Stabilizedhover Pilotoutbrief11/30GrandCasinoFlight112/2HardRockFlight1 Dualightmodes Post-deploymentsynthesis Videoandhi-resstills InterviewwithFEMAUSARspecialist[ 40 ] Obstacleavoidance Pilotoutbrief11/30HardRockFlight212/1PresidentBargeFlight412/21550BeachBlvdFlights1,312/3PresidentBargeFlight3 Upgradedradio Pilotoutbrief11/30HardRockFlight212/1PresidentBargeFlights1-4 Payloadstabilization MissionSpecialistoutbrief11/30HardRockFlight112/2CasinoMagicFlight1 Payload`return-to-center' MissionSpecialistoutbrief11/30HardRockFlight1 `Hoverhere' Pilotoutbrief11/30GrandCasinoFlight112/21550BeachBlvd.Flight2Post-deploymentsynthesis 42

PAGE 53

Chapter8ConclusionTheconclusiontothisworkbreaksdownintotwocomponents,Section 8.1 whichsumarizestheresultsdiscussedthusfar,andSection 8.2 whichlaysoutadevelopmentpathforvehicleautonomyin VTOL MAV s.8.1SummaryWhenitassaultedtheGulfCoastinthelatesummerof2005,HurricaneKatrinainstan-taneouslybecameoneofthemostdisastroushurricanesinrecentmemory;ahurricanewhoseeffectswillcontinuetobefeltintheGulfCoastregionforyears,andlikelydecades,tocome.Inadditiontothistragicdevastation,HurricaneKatrinaisofhistoricalimportforanotherreason,namely,theresponseandrecoveryphasesofHurricaneKatrinasawtherstoperationalusageof UAV sduringadisasterresponse. CRASAR deployedteamsequippedwith VTOL MAV stwicetothedisasterarea;onceimmediatelyfollowingthehurricaneduringtheresponsephase,andagain90dayslaterduringtherecoveryphase.Asthesearetherstoperationaldeploymentsof MAV s,therearemanyquestionstheycanhelpanswer,butthethreecoreresearchquestionsareasfollows:Whatteamprocessesarenecessarytoyan MAV inaclutteredurbanenvironment?,Whattechnicalcapabilitiesarerequiredtousean MAV platforminthisenvironment?,andWhatisthemosteffectivepathtoevolve MAV sfromsemi-autonomoustofullyautonomous?Usingthesequestionsasastartingpoint,videologsandcrewdebriengswereanalyzedtoarriveatthefollowingfourcentralconclusions.First,withbasic COTS opticstheminimumnecessarystandoff 43

PAGE 54

rangefora VTOL MAV inastructuralinspectiontaskis2-5m.Second,whileitismostintuitivetoimagetargetsstraightaheadatadownelevation,obstaclescanandwillen-croachonthevehicle'sairspacefromallangles;an MAV inurbanenvironmentsneedsomni-directionalobstacleavoidancetosuccessfullymovetohigherlevelsofautonomy.Third,giventheoperationaltempoandthehighlydynamicnatureoftheights,itdoesnotappearthatGPSwaypointnavigationwouldbeausefulfeatureforurbaninspectiontypeoperations.Andnally,safetyandinformationalcontextconcernsdictatethatforthetimebeingtheseoperationstakethreeoperatorstorunone MAV intheeld.8.2FutureWorkSowhatofthewayforwardthen?Themostpressingresearchareasforfutureworkareinincrementalautonomyandabetterunderstandingofthehuman-factorsandinterfaceissues.Thisincludesareassuchasplatformstabilization,payloadstabilizationandvisualservo-ing,completesphericalobstacleavoidance,augmentedrealitypilotdisplays,multimodalinterfaces,andteamperformancemetrics.Withsuchalisthoweveritisimportanttoprovideamoreformalizedanddirectedversion,somethingwhichcandirectandguidefutureresearchin VTOL MAV autonomy.Figure 23 illustratestheproposeddevelopmentplan.TheupperleftcornerofFigure 23 showsthattheoperationsconductedfollowingHur-ricaneKatrinawereentirelyteleoperated.Thenextstepfromdirectteleoperationpilotcommandsasactuatorinputs,isattitudestabilizationoray-by-wiredesignpilotinputsasrequested6-DOFattitudemodications.InthetimesinceKatrina,someplatformshavereachedthisattitudestabilizationcapability,withmostcurrentplatformsoperatingsomewherealongthecontinuumbetweenthesestates.Thenextsteppaststabilizationisobstacleavoidanceandaguardedmotionbehavior,whereinthepilotrequestedattitudead-justmentsarenowlteredwithanobstacleavoidancebehaviortopreventcollisions.The 44

PAGE 55

Figure23:DevelopmentpathfortheimplementationofautonomyinVTOLMAVsforurbaninspectionapplications. naltargetstateforautonomydevelopmentin VTOL MAV sshouldbepayloaddirectedight.Fieldoperatorsofsuchaplatform,suchas USAR operators,areonlyinterestedinthepayloaddataoutput,nothowthevehiclestaysintheair.Thus,ifthepilotoroperatorcansimplydirectthepayloadcameratowardstheirintendedtarget,andthehelicopterusesitsstabilizedhoverandobstacleavoidancebehaviorstofollowthepayloadastheoperatorysthecamera,operationandrequiredtrainingforsuchaplatformwouldbothbegreatlysimplied.AlthoughFigure 23 presentsthisasalinearprogression,thetwolayerbreakdownisnotentirelyaccidental.Thersttwocontrolmethodologiescanalsobethoughtofassimplyteleoperationandadvancedteleoperation;theattitudestabilizationofoadssomeoftheinnerloopcontrolrequirementsfromthepilot,butitdoesnotmakeanycontrolinputstodrivetheaircrafttoanyparticularlocation.Droppingdowntolayertwoandobstacleavoidancehowevertheautonomybegainstoaidthepilotinmoving,ormoreoftennotmoving,theaircraftthroughtheenvironment.Withthenalstepofpayloaddirectedight,theautomationbecomesresponsibleforallofthedriveinputs,respondingonlyto 45

PAGE 56

qualitativepilotcommandssuchas`upandtotheleft'.Withthisprogressionupthroughthelayersofthepathplanningsystem,thereislittleoptionforthisprogressiontohappeninanyothermanner.Asanalpoint,itisimportanttocomparethisproposeddevelopmentpathwithwhatappearstobethemoretraditionalGPS-baseddevelopmentpath.Fortunatelythisdifferenceappearstobasedononecoredifferencebetweenthemethodologies:frameofreference.Theaboveproposeddevelopmentpathorientsthecontrolandautonomyproblemstothepilot'sframeofreference,whiletraditionalGPSsystemsarederivedfromaworldbasedabsolutecoordinatesystem;itisaquestionofhuman-centereddesigncomparedtoaworldreferencedmethodology.Inthetargetstateofthepropsedsystemapilotdesignatesatargetbysimplypointingthecameraatit,whereasdesignatingatargetinaGPSsystemoftenrequiresenteringalat.long.coordinatepairforthetarget.Whilegeo-referencedcoordinatesystemsworkwellforlarger UAV soperatingonaglobalscale,smaller MAV soperatingatalocalandhumanscalearebetterservedbyoperatingwithinthiscoordinatesystem. 46

PAGE 57

References [1] I.A.WilliamEvans,R.M.Hoeft,F.Jentsch,S.A.Rehfeld,andM.T.Curtis.Explor-inghuman-robotinteraction:Emergingmethodologiesandenvironments.InN.J.Cooke,H.L.Pringle,H.K.Pederson,andO.Connor,editors,HumanFactorsofRemotelyOperatedVehicles,volume7,pages345.JAI,2006. [2] M.R.Blackburn,H.R.Everett,andR.T.Laird.Afteractionreporttothejointpro-gramofce:Centerfortheroboticassistedsearchandrescuecrasarrelatedeffortsattheworldtradecenter.Technicalreport,SPAWARSystemsCenterSanDiego,August2002. [3] G.Bland,P.Coronado,T.Miles,andJ.P.Bretthauer.TheAEROSproject:Experi-mentswithsmallelectricpowereduavsforearthscience.InInfotech@Aerospace,Arlington,VA,26-29September2005. [4] J.L.Burke.Moonlightinmiami:Aeldstudyofhuman-robotinteractioninthecontextofanurbansearchandrescuedisasterresponsetrainingexercise.Master'sthesis,UniversityofSouthFlorida,4202E.FowlerAve.Tampa,FL33620,September2003. [5] G.L.CalhounandM.H.Draper.Multi-sensoryinterfacesforremotelyoperatedvehicles.InN.J.Cooke,H.L.Pringle,H.K.Pederson,andO.Connor,editors,HumanFactorsofRemotelyOperatedVehicles,volume7,pages149.JAI,2006. [6] J.Casper.Human-robotinteractionsduringtherobot-assistedurbansearchandrescueresponseattheworldtradecenter.Master'sthesis,UniversityofSouthFlorida,4202E.FowlerAve.Tampa,FL33620,May2002. [7] R.A.Chadwick,S.L.Pazuchanics,andD.J.Gillan.Whattherobot'scameratellstheoperator'sbrain.InN.J.Cooke,H.L.Pringle,H.K.Pederson,andO.Connor,editors,HumanFactorsofRemotelyOperatedVehicles,volume7,pages373.JAI,2006. [8] W.J.Clancey.ObservationofWorkPracticesinNaturalSettings,pages127.CambridgeUniversityPress,2002. [9] A.Gage,R.R.Murphy,E.Rasmussen,andB.Minten.Shadowbowl2003[simulatedmass-casualtyexercise].IEEERobotics&AutomationMagazine,11:62,Sept.2004. 47

PAGE 58

[10] R.A.Garay.Marinecorpssystemscommandliaisonteam:Fieldreportcentraliraq,20aprilto25april2003.Fieldreport,UnitedStatesMarineCorps,2003. [11] S.Gourley.Ravenuav.ArmyMagazine,55,2005. [12] W.GreenandP.Oh.Autonomoushoveringofaxed-wingmicroairvehicle.InRoboticsandAutomation,2006.ICRA2006.Proceedings2006IEEEInternationalConferenceon,pages2164,May15-19,2006. [13] D.E.Grilley.Resolutionrequirementsforpassivesenseandavoid.Technicalreport,AlionScienceandTechnology,3592CollinsFerryRd.Ste.180,Morgantown,WV,26505,January262005. [14] G.J.Holland,P.J.Webster,J.A.Curry,G.Tyrell,D.Gauntlett,G.Brett,J.Becker,R.Hoag,andW.Vaglienti.Theaerosonderoboticaircraft:Anewparadigmforenvi-ronmentalobservations.BulletinoftheAmericanMeteorologicalSociety,82:889902,2001. [15] S.B.HottmanandK.Sortland.Uavoperators,otherairspaceusers,andregulators:Criticalcomponentsofanuninhabitedsystem.InN.J.Cooke,H.L.Pringle,H.K.Pederson,andO.Connor,editors,HumanFactorsofRemotelyOperatedVehicles,volume7,pages71.JAI,2006. [16] S.Hrabar,G.Sukhatme,P.Corke,K.Usher,andJ.Roberts.Combinedoptic-owandstereo-basednavigationofurbancanyonsforauav.InIntelligentRobotsandSystems,2005.IROS2005.2005IEEE/RSJInternationalConferenceon,pages3309,2-6August2005. [17] B.P.Hunn.Videoimagery'sroleinnetworkcentric,multipleunmannedaerialvehi-clesuavoperations.InN.J.Cooke,H.L.Pringle,H.K.Pederson,andO.Connor,editors,HumanFactorsofRemotelyOperatedVehicles,volume7,pages179.JAI,2006. [18] B.P.HunnandO.H.Heuckeroth.Ashadowunmannedaerialvehicleuavimprovedperformanceresearchintergrationtoolimprintmodelsupportingfuturecombatsys-tems.TechnicalReportARL-TR-3731,ArmyResearchLaboratory,February2006. [19] H.L.Jones,E.W.Frew,B.R.Woodley,andS.M.Rock.Human-robotinteractionforeldoperationofanautonomoushelicopter.InH.M.Choset,editor,MobileRobotsXIIIandIntelligentTransportationSystems,volume3525.SPIE,1998. [20] R.D.Knabb,J.R.Rhome,andD.P.Brown.Tropicalcyclonereport:Hurricanekatrina:23-30august2005.Technicalreport,NationalHurricaneCenter,August2006. 48

PAGE 59

[21] J.S.McCarleyandC.D.Wickens.Humanfactorsimplicationsofuavsinthenationalairspace.TechnicalReportAHFD-05-5/FAA-05-1,UniversityofIllinoisInstituteofAviation:AviationHumanFactorsDivision,Savoy,IL,2005. [22] T.McGeerandJ.Vagners.Wide-scaleuseoflong-rangeminiatureaerosondesovertheworld'soceans.InProceedingsofAUVSI26thAnnualSymposium.AssociationforUnmannedVehiclesSystemsInternational,July1999. [23] V.M.McHuch,B.S.Ince,G.E.Blethen,C.S.Harden,R.J.Schafer,S.E.Harper,P.D.Arnold,andM.A.Thomas.Updateonanunmannedaerialvehicleuavpay-loadfordetection,identication,andacquisitionofvaporsoftoxicsubstancesandtheirprecursors.InISIMS200413thInternationalConferenceonIonMobilitySpec-tometry,2004. [24] M.Micire.Analysisoftherobotic-assistedsearchandrescueresponsetotheworldtradecenterdisaster.Master'sthesis,UniversityofSouthFlorida,4202E.FowlerAve.Tampa,FL33620,July2002. [25] R.R.Murphy.Human-robotinteractioninrescuerobotics.Systems,ManandCy-bernetics,PartC:ApplicationsandReviews,IEEETransactionson,34:138,May2004. [26] R.R.MurphyandJ.L.Burke.Upfromtherubble:Lessonslearnedabouthrifromsearchandrescue.InProceedingsofthe49thAnnualMeetingoftheHumanFactorsandErgonomicsSociety,September2005. [27] NGA.Comprehensiveemergencymanagement:Agovernorsguide.Technicalreport,NationalGovernors'Association,1979. [28] T.Oron-Gilad,J.Y.C.Chen,andP.A.Hancock.Remotelyoperatedvehiclesrovsfromthetop-downandfromthebottom-up.InN.J.Cooke,H.L.Pringle,H.K.Pederson,andO.Connor,editors,HumanFactorsofRemotelyOperatedVehicles,volume7,pages37.JAI,2006. [29] H.K.Pederson,N.J.Cooke,H.L.Pringle,andO.Connor.Uavhumanfactors:Operatorperspectives.InN.J.Cooke,H.L.Pringle,H.K.Pederson,andO.Connor,editors,HumanFactorsofRemotelyOperatedVehicles,volume7,pages21.JAI,2006. [30] K.S.Pratt,R.R.Murphy,S.Stover,andC.Grifn.Requirementsforsemi-autonomousightinminiatureuavsforstructuralinspection.InProceedingsofAu-tonomousUnmannedVehiclesSytemsInternationalAUVSI2006,2006. [31] M.Quigley,B.Barber,S.Grifths,andM.A.Goodrich.Towardsreal-worldsearch-ingwithxed-wingmini-uavs.InIntelligentRobotsandSystems,2005.IROS2005.2005IEEE/RSJ,Edmonton,Alberta,Canada,August2005. 49

PAGE 60

[32] M.Quigley,M.A.Goodrich,andR.W.Beard.Semi-autonomoushuman-uavinter-facesforxed-wingmini-uavs.InIntelligentRobotsandSystems,2004.IROS2004.2004IEEE/RSJInternationalConferenceon,Sendai,Japan,September28-October22004. [33] M.Quigley,M.A.Goodrich,S.Grifths,A.Eldgredge,andR.W.Beard.Targetacquisition,localization,andsurveillanceusingaxed-wingmini-uavandgimbaledcamera.InRoboticsandAutomation,2005.ICRA2005.Proceedings2005IEEEInternationalConferenceon,Barcelona,Spain,April2005. [34] M.A.Raffetto.Unmannedaerialvehiclecontributionstointelligence,surveillance,andreconnaissancemissionsforexpeditionaryoperations.Master'sthesis,NavalPostrgraduateSchool,Monterey,CA,September2004. [35] N.Sabatini.Unmannedaircraftoperationsinthenationalairspacesystem.FAANotice,February62007. [36] S.Scherer,S.Singh,L.Chamberlain,andS.Saripalli.Flyingfastandlowamongobstacles.InRoboticsandAutomation,2007.ICRA2007.Proceedings2007IEEEInternationalConferenceon,2007. [37] W.Schneider.Defensescienceboardstudyonunmannedaerialvehiclesandunin-habitedcombataerialvehicles.Technicalreport,OfceoftheUnderSecretaryofDefenceForAcquisition,Technology,andLogistics,February2004. [38] B.P.Self,W.R.Ercoline,W.A.Olson,andA.P.Tvaryanas.Spatialdisorientationinuninhabitedaerialvehicles.InN.J.Cooke,H.L.Pringle,H.K.Pederson,andO.Connor,editors,HumanFactorsofRemotelyOperatedVehicles,volume7,pages133.JAI,2006. [39] D.H.Shim,H.Chung,H.J.Kim,andS.Sastry.Autonomousexplorationinun-knownurbanenvironmentsforunmannedaerialvehicles.InProceedingsoftheAIAAConferenceonGuidance,Navigation,andControl,2005.AIAAGN&C2005.,2005. [40] S.Stover.PersonalInterviewwithSamStover,memberIN-TF1,December2005. [41] R.J.V.Vuren.Advisorycircular91-57.FAANotice,June91981. [42] R.E.WeibelandR.J.Hansman.Safetyconsiderationsforoperationofdifferentclassesofuavsinthenas.InAIAA4thAviationTechnology,Integration,andOpera-tionsATIOForum.AIAA,2004. [43] R.E.WeibelandR.J.Hansman.Safetyconsiderationsforoperationofunmannedaerialvehiclesinthenationalairspacesystem.TechnicalReportICAT-2005-1,MITInternationalCenterforAirTransportation,DepartmentofAeronauticsandAstronau-tics,MassachusettsInstituteofTechnology,Cambridge,MA,02139,March2005. 50

PAGE 61

[44] J.-C.ZuffereyandD.Floreano.Toward30-gramautonomousindooraircraft:Vision-basedobstacleavoidanceandaltitudecontrol.InRoboticsandAutomation,2005.ICRA2005.Proceedings,2005. 51