预览加载中,请您耐心等待几秒...
1/3
2/3
3/3

在线预览结束,喜欢就下载吧,查找使用更方便

如果您无法下载资料,请参考说明:

1、部分资料下载需要金币,请确保您的账户上有足够的金币

2、已购买过的文档,再次下载不重复扣费

3、资料包下载后请先用软件解压,在使用对应软件打开

局限空间中基于单目视觉的四旋翼飞行器的位姿预估与导航 Title:PoseEstimationandNavigationofQuadrotorAircraftBasedonMonocularVisioninConfinedSpaces Abstract: Quadrotoraircrafthavegainedsignificantattentioninrecentyearsduetotheirabilitytonavigateinconfinedspaces.However,accuratelyestimatingthepose(positionandorientation)ofaquadrotorinsuchenvironmentsisachallengingtask.Thispaperaimstoexploretheuseofmonocularvisionforposeestimationandnavigationofquadrotoraircraftinconfinedspaces.Wediscussthelimitationsofthecurrentstate-of-the-arttechniquesandproposeanovelapproach. 1.Introduction Quadrotoraircraftareincreasinglybeingdeployedinvariousapplications,includingsurveillance,inspection,andsearchandrescuemissions.However,thesetasksoftenrequirethequadrotortooperateinconfinedspaces,suchasindoorenvironmentsorclutteredoutdoorareas.Accurateposeestimationandnavigationbecomecrucialforsafeandefficientoperationinsuchscenarios. 2.RelatedWork Previousresearchhasexploreddifferentmethodsforposeestimationandnavigationofquadrotoraircraft.Theseincludetheuseofsensorssuchasinertialmeasurementunits(IMUs),globalpositioningsystem(GPS),andvision-basedapproaches.Vision-basedmethodshavethepotentialtoprovideaccurateandrobustposeestimationinconfinedspaces. 3.ChallengesinConfinedSpaces Confinedspacesposeseveralchallengesforquadrotornavigation,includinglimitedvisualfeatures,occlusions,dynamicenvironments,andlackofabsolutepositioningreferences.Thesechallengesmakeitdifficulttoestimatethequadrotor'sposeaccuratelyandreliably. 4.MonocularVision-basedPoseEstimation Monocularvisionreferstotheuseofasinglecameraforposeestimation.Weproposeanovelapproachthatleveragesbothfeature-basedanddirectmethodsforposeestimation.Feature-basedmethodsuseextractedvisualfeatures,suchascorners,edges,orkeypoints,toestimatethepose.Directmethods,ontheotherhand,estimatetheposedirectlyfromthepixelintensityinformation. 5.PoseEstimationAlgorithms Wepresentacomprehensiveoverviewofexistingposeestimationalgorithmsandtechniques,includingvisualodometry,simultaneouslocalizatio