欢迎来到天天文库
浏览记录
ID:34574624
大小:94.06 KB
页数:48页
时间:2019-03-08
《Lecture for Kalman filter.pdf》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、STCLectureSeriesAnIntroductiontotheKalmanFilterGregWelchandGaryBishopUniversityofNorthCarolinaatChapelHillDepartmentofComputerSciencehttp://www.cs.unc.edu/~welch/kalmanLinks.htmlUNCChapelHillComputerScienceSlide1WhereWe’reGoing•Introduction&Intuition•TheDiscreteKalmanFilter•ASimpleExample•Variation
2、softheFilter•RelevantApplications&ReferencesUNCChapelHillComputerScienceSlide2TheKalmanfilter•SeminalpaperbyR.E.Kalman,1960•Setofmathematicalequations•Optimalestimator–minimummeansquareerror•Versatile–Estimationpredictpredictcorrectcorrect–Filtering–Prediction–FusionUNCChapelHillComputerScienceSlid
3、e3WhyaKalmanFilter?•Efficient“least-squares”implementation•Past,presentandfutureestimation•Estimationofmissingstates•Measureofestimationquality(variance)•Robust–forgivinginmanyways–stablegivencommonconditionsUNCChapelHillComputerScienceSlide4SomeIntuitionUNCChapelHillComputerScienceSlide5FirstEstim
4、ateConditionalDensityFunction2z1,sz1xˆ=z11N(z,sz2)1122sˆ1=sz1-202468101214UNCChapelHillComputerScienceSlide6SecondEstimateConditionalDensityFunction2z2,sz22N(z,sz)2xˆ=...?222sˆ2=...?-202468101214UNCChapelHillComputerScienceSlide7CombineEstimates222222xˆ2=[]sz2()sz1+sz2z1+[]sz1()sz1+sz2z2=xˆ1+K2[]z2
5、-xˆ1where222K2=sz()sz+sz112UNCChapelHillComputerScienceSlide8CombineVariances2221s2=()1sz+()1sz12UNCChapelHillComputerScienceSlide9CombinedEstimateDensityConditionalDensityFunctionN(x,ˆsˆ2)xˆ=xˆ222sˆ=s2-202468101214UNCChapelHillComputerScienceSlide10AddDynamicsdx/dt=v+wwherevisthenominalvelocitywis
6、anoiseterm(uncertainty)UNCChapelHillComputerScienceSlide11PropagationofDensityUNCChapelHillComputerScienceSlide12SomeDetailsxxA=+AxxwwzzH=Hxx........UNCChapelHillComputerScienceSlide13DiscreteKalmanFilterMaintainsfirsttwostatisticalmomentsprocessstate(mean)zyxerrorcovarianceUNCChapelHillComputerSci
7、enceSlide14DiscreteKalmanFilterTheIngredients•Adiscreteprocessmodel–changeinstateovertime–lineardifferenceequation•Adiscretemeasurementmodel–relationshipbetweenstateandmeasurement–linearfunction•ModelParame
此文档下载收益归作者所有