Self-Driving Car ND - Sensor Fusion - Extended Kalman Filters

Size: px
Start display at page:

Download "Self-Driving Car ND - Sensor Fusion - Extended Kalman Filters"

Transcription

1 Self-Driving Car ND - Sensor Fusion - Extended Kalman Filters Udacity and Mercedes February 7, 07 Introduction Lesson Ma 3 Estimation Problem Refresh 4 Measurement Udate Quiz 5 Kalman Filter Equations In C++ The state transition function is x f(x) + ν F x + }{{} Bu +ν () 0 The motion control, Bu, is zero in our case, so our state transition function is simlified to: The measurement function is x F x + ν () z h(x ) + ω Hx + ω (3) where, ν is the rocess noise. It is a Gaussian with zero mean and covariance matrix Q: ν N(0, Q) (4) ω is the measurement noise. It is a Gaussian with zero mean and covariance matrix R: ω N(0, R) (5) The state vector: Linear motion: x ( ) v (6) + v t (7) v v (8)

2 The state rediction function: The measurement function: ( ) v ( t 0 ) ( v z ( 0 ) ( v ) ) (9) (0) 5. Kalman Filter Algorithm Prediction: x F x + u () P F P F T + Q () Measurement Udate: y z Hx (3) S HP H T + R (4) K P H T S (5) x x + Ky (6) P (I KH)P (7) 6 Kalman Filter Equation in C++ Programming Assignment 7 Measurement Udate Quiz 8 State Prediction - 4D Case x y v x (8) 9 Uncertainty Deending on Time Delay Quiz 0 Uncertainty Deending on Acceleration Quiz

3 Process Covariance Matrix Linear Motion Model - D: Can be exressed in a matrix form as: x 0 t 0 y v x 0 0 t y v x x F x + ν (9) x + v x t + ν x y y + t + ν y v x (0) v x + ν vx + ν vy. Deriving the D velocity with a constant acceleration ν x ν y ν vx ν vy () From the kinematic formulas we can define the acceleration as the change in the velocity over the elased time, so the current velocity minus revious velocity over the elased time. a v t v k+ v k () t Then we can derive the current velocity as the revious velocity lus the acceleration a times t. v k+ v k + a t (3) Generalized in D we have both seed comonents described as follows: { v x v x + a x t v y + a y t. Deriving the relation between the revious location and current osition in D when having a motion with constant acceleration Similarly we know that the dislacement equals the average velocity v avg times t. (4) k+ k v avg t (5) which is the revious velocity v k+ lus current velocity v k over two. k+ k v k+ + v k t (6) k+ k + v k+ + v k t (7) And as we already saw, the current velocity v k+ is our old velocity v k lus the acceleration a times t. v k+ v k + a t (8) so we can relace the v k+ with v k + a t in revious equations: k+ k + v k + a t + v k t (9) k+ k + v k t + a t (30) And finally we get the full relation between the revious location k and current location k+ : k+ k + v k t + a t Going in D, we ll have both horizontal and vertical directions: { x + v x t + ax t y y + t + ay t (3) (3) 3

4 .3 The D motion model with constant acceleration Combining both D osition and D velocity equations reviously deducted formulas we have: x + v x t + ax t y y + t + ay t (33) v x v x + a x t + a y t.4 Process Noise in D Since the acceleration is unknown we can add it to the noise comonent, and this random noise would be exressed analytically as the last terms in the equation derived above. So, we have a random acceleration vector ν in this form: ν ν x ν y ν vx ν vy a x t a y t a x t a y t which is described by a zero mean and a covariance matrix Q. (34) ν N(0, Q) (35) The vector ν can be decomosed into two comonents a 4 by matrix G which does not contain random variables and a by matrix a which contains the random acceleration comonents: a x t t a y t ν 0 ( ) a x t t 0 ax t 0 Ga (36) a y }{{} a y t 0 t a }{{} G t is comuted at each Kalman Filter ste and the acceleration is a random vector with zero mean and standard deviations σ ax and σ ay. Based on our noise vector we can define now the new covariance matrix Q. The covariance matrix is defined as the exectation value of the noise vector ν times the noise vector ν transose. So let s write this down: Q E[νν T ] E[Gaa T G T ] (37) As G does not contain random variables, we can ut it outside the exectation calculation. ( ) σ Q GE[aa T ]G T G ax σ axy G T GQ ν G T (38) This leaves us with three statistical moments: σ axy the exectation of ax times ax, which is the variance of ax: σ ax. the exectation of ay times ay, which is the variance of ay: σ ay squared. And the exectation of ax times ay, which is the covariance of ax and ay: σ axy. a x and a y are assumed uncorrelated noise rocesses. This means that the covariance σ axy in Q ν is zero: ( ) ( ) σ Q ν ax σ axy σ σ axy σay ax 0 0 σay (39) So after combining everything in one matrix we obtain our 4 by 4 Q matrix: Q GQ ν G T σ ay t 4 4 σ t ax 0 3 t 0 4 σ ax 0 4 σ t ay 0 3 σ ay t 3 σ ax 0 t σax 0 0 t 3 σ ay 0 t σay 4 (40)

5 .5 other equations Note: Some authors describe Q as the comlete rocess noise covariance matrix. And some authors describe Q as the covariance matrix of the individual noise rocesses. In our case, the covariance matrix of the individual noise rocesses matrix is called Q ν. So we should be aware of that. Process Covariance Matrix Text 3 Laser Measurements 4 Find Out the H Matrix 4. Quiz question 4. Quiz answer ( x ( x ) y y ) (? ) x y v x (4) ( ) x y v x (4) 5 Find out the Dimension of R 5. Quiz question Now, can you determine what is the dimensionality of the measurement noise covariance matrix R. 5. Quiz answer ( ) σ R E[ωω T ] x 0 0 σy (43) 6 Programming Assignment ( ) 0. 0 R 0 0. ( ) 0 R 0 (44) (45) 6. Other Equations a x N(0, σ ax) (46) a y N(0, σ ay) (47) 5

6 7 Radar Measurements The state transition function: The measurement function: x f(x) + ν (48) z h(x ) + ω (49) The state transition function will be exactly what you ve designed in the lidar case. However our new Radar sees the world differently. Instead of a D ose (, y ), the radar can directly measure the object range ρ, bearing ϕ and range rate ρ: ρ z ϕ (50) ρ The range ρ is the radial distance from the origin to our edestrian. So we can always define a ray which extends from the origin to our object osition. The bearing ϕ is the angle between the ray and x direction. And the range rate ρ, also known as Doler or radial velocity is the velocity along this ray. And similar to our motion model, the radar observations are corruted by a zero-mean random noise ω N(0, R). Considering that the three measurement comonents of the measurement vector z are not cross-correlated, the radar measurement covariance matrix R becomes a 3 by 3 diagonal matrix: σ ρ 0 0 R 0 σϕ 0 (5) 0 0 σ ρ And the next question is what is the measurement function h(x ) that mas the redicted state x into the measurement sace? ρ x h(x ϕ ) y v x ρ (5) And that s how I defined the h function which basically secifies how the redicted osition and seed can be related to the observed range ρ, bearing ϕ and range rate ρ: x + y h(x ) arctan( y/ x) (53) x v x + y x + y 7. Deriving the Radar Measurement Function The measurement function is comosed of three comonents which show how the redicted state x ( x, y, v x, v y) T is maed into the measurement sace z (ρ, ϕ, ρ) T :. The range ρ is the distance to the edestrian which can be defined as: ρ x + y (54). ϕ is the angle between ρ and x direction and can be defined as: 3. Range rate ρ(t) derivation: ϕ arctan( y / ) (55) 6

7 Method : Generally we can exlicitly describe the range ρ as a function of time: ρ(t) (t) + y (t) (56) The range rate ρ(t) is defined as time rate of change of the range ρ, and it can be described as the time derivative of ρ: ρ (t) t t (t) + y (t) (t) + y (t) ( t (t) + t y(t) ) (t) + y (t) ((t) t (t) + y (t) t y(t)) t (t) is nothing else than v x (t), similarly t y(t) is (t). So we have: ρ (t) t (t) + y (t) ((t)v x (t)+ y (t) (t)) ((t)v x (t) + y (t) (t)) (t) + y (t) (t)v x (t) + y (t) (t) x (t) + y (t) For the simlicity we just use the following notation: ρ v x + y v y x + y (57) Method : The range rate ρ can be seen as a scalar rojection of the velocity vector v onto ρ. Both ρ and v are D vectors defined as: ( ) ( ) x vx ρ, v (58) y The scalar rojection of the velocity vector v onto ρ is defined as: ( ) ( ) vx x ρ v ρ ρ y v x + y v y x + y x + y (59) where ρ is the length of ρ. In our case it is actually the range, so ρ ρ 8 Maing with a Nonlinear Function Quiz 9 Extended Kalman Filter 0 Linearization Examle 0.0. Linearization of a function The Extended Kalman Filter uses a method called first order Taylor exansion h(x) h(µ) + h(µ) (x µ) (60) x 7

8 0. 0_quiz_linearization_examle 0.. quiz answer In our examle µ 0, therefore: h(x) x h(x) arctan(x) (6) arctan(x) x + x (6) h(x) arctan(x) (63) h(x) h(µ) + h(µ) (x µ) arctan(µ) + (x µ) (64) x + µ h(x) arctan(0) + (x 0) x (65) + 0 So, the function h(x) arctan(x) will be aroximated by a line: Jacobian Matrix State transition function: Measurement function: h(x) x (66) x f(x) + ν (67) z h(x ) + ω (68) where we consider that f(x) and h(x) are nonlinear and can be linearized as: f(x) f(µ) + f(µ) (x µ) (69) } x {{} F j h(x) h(µ) + h(µ) (x µ) (70) } x {{} H j The derivative of f(x) and h(x) with resect to x are called Jacobians: f(µ) x F j (7) h(µ) x H j (7) and it is going to be a matrix containing all the artial derivatives: H j h h x h h x. h m x h x n h x n x... x h m x... To be more secific, I know that my function h describes three comonents: range ρ, bearing ϕ and range-rate ρ, and my state is a vector with four comonents, y, v x,. Then the Jacobian matrix H j is going to be a matrix with 3 rows and four columns: H j y y y h m x n (73) (74) 8

9 After comuting all these artial derivatives we have: H j y x + y x + y y x + y x + y y(v x y ) ( v x y) ( x + y )3/ ( x + y )3/ x + y y x + y (75) If you re interested to see more details about how I got this results have a look at details below..0. Derivation of the Jacobian H j We re going to calculate, ste by ste, all the artial derivatives in H j : H j y y y So all the matrix H j elements are calculated as follows: (76). ( x + y) x x x + y x + y (77). y ( x + y) y y x + y y x + y (78) ( v x + y) 0 (79) x ( v x + y) 0 (80) y arctan( y / ) arctan( y / ) y y ( y ( y ) + ( y x ) y x + y ) + ( ) x + y x + y (8) (8) arctan( y / ) 0 (83) arctan( y / ) 0 (84) v x + y x + y (85) In order to calculate the derivative of this function we use the quotient rule. Given a function z that is quotient of two other functions f and g: z f g (86) 9

10 its derivative with resect to x is defined as: in our case: Their derivatives are: f z x x g g x f g (87) f v x + y (88) g x + y (89) f ( v x + y ) v x (90) g ( ) x + y x x + y (9) Putting everything together into the derivative quotient rule we have: v x x + y x ( x v x + y ) x + y x + y(v x y ) (9) y ( x + y) 3/ 0. Similarly y y v x + y x + y ( v x y ) ( x + y) 3/ (93).. v x + y x + y v x + y x + y x + y y x + y (94) (95) So now, after calculating all the artial derivatives our resulted Jacobian H j is: y 0 0 x + y x + y H j y x + y 0 0 x + y y(v x y ) ( v x y) ( x + y )3/ ( x + y )3/ EKF Algorithm Generalization 3 EKF with Linear Functions Quiz x + y 4 Sensor Fusion General Processing Flow 5 General EKF Algorithm y x + y (96) 0

11 6 Evaluating the Performance 6. RMSE RMSE n n t (x est t x true t ) (97) 7 3_State rediction - 4D case // T_SC_ Linear Motion - D: State transition: Prediction: Measurement Udate: x y v x }{{} x x + v x t y y + t v x v x 0 t t } 0 0 {{ } F y v x }{{} x (98) (99) x f(x, u) (00) P F j P F T j + Q (0) y z h(x ) (0) 8 Outro z (03) z ( x y ) (04) References

1 Kalman Filter Introduction

1 Kalman Filter Introduction 1 Kalman Filter Introduction You should first read Chapter 1 of Stochastic models, estimation, and control: Volume 1 by Peter S. Maybec (available here). 1.1 Explanation of Equations (1-3) and (1-4) Equation

More information

General Random Variables

General Random Variables Chater General Random Variables. Law of a Random Variable Thus far we have considered onl random variables whose domain and range are discrete. We now consider a general random variable X! defined on the

More information

FE FORMULATIONS FOR PLASTICITY

FE FORMULATIONS FOR PLASTICITY G These slides are designed based on the book: Finite Elements in Plasticity Theory and Practice, D.R.J. Owen and E. Hinton, 1970, Pineridge Press Ltd., Swansea, UK. 1 Course Content: A INTRODUCTION AND

More information

Kalman Filter. Predict: Update: x k k 1 = F k x k 1 k 1 + B k u k P k k 1 = F k P k 1 k 1 F T k + Q

Kalman Filter. Predict: Update: x k k 1 = F k x k 1 k 1 + B k u k P k k 1 = F k P k 1 k 1 F T k + Q Kalman Filter Kalman Filter Predict: x k k 1 = F k x k 1 k 1 + B k u k P k k 1 = F k P k 1 k 1 F T k + Q Update: K = P k k 1 Hk T (H k P k k 1 Hk T + R) 1 x k k = x k k 1 + K(z k H k x k k 1 ) P k k =(I

More information

EE 508 Lecture 13. Statistical Characterization of Filter Characteristics

EE 508 Lecture 13. Statistical Characterization of Filter Characteristics EE 508 Lecture 3 Statistical Characterization of Filter Characteristics Comonents used to build filters are not recisely redictable L C Temerature Variations Manufacturing Variations Aging Model variations

More information

A Time-Varying Threshold STAR Model of Unemployment

A Time-Varying Threshold STAR Model of Unemployment A Time-Varying Threshold STAR Model of Unemloyment michael dueker a michael owyang b martin sola c,d a Russell Investments b Federal Reserve Bank of St. Louis c Deartamento de Economia, Universidad Torcuato

More information

Miscellaneous. Regarding reading materials. Again, ask questions (if you have) and ask them earlier

Miscellaneous. Regarding reading materials. Again, ask questions (if you have) and ask them earlier Miscellaneous Regarding reading materials Reading materials will be provided as needed If no assigned reading, it means I think the material from class is sufficient Should be enough for you to do your

More information

MULTI-CHANNEL PARAMETRIC ESTIMATOR FAST BLOCK MATRIX INVERSES

MULTI-CHANNEL PARAMETRIC ESTIMATOR FAST BLOCK MATRIX INVERSES MULTI-CANNEL ARAMETRIC ESTIMATOR FAST BLOCK MATRIX INVERSES S Lawrence Marle Jr School of Electrical Engineering and Comuter Science Oregon State University Corvallis, OR 97331 Marle@eecsoregonstateedu

More information

State Estimation with ARMarkov Models

State Estimation with ARMarkov Models Deartment of Mechanical and Aerosace Engineering Technical Reort No. 3046, October 1998. Princeton University, Princeton, NJ. State Estimation with ARMarkov Models Ryoung K. Lim 1 Columbia University,

More information

Bayes Filter Reminder. Kalman Filter Localization. Properties of Gaussians. Gaussians. Prediction. Correction. σ 2. Univariate. 1 2πσ e.

Bayes Filter Reminder. Kalman Filter Localization. Properties of Gaussians. Gaussians. Prediction. Correction. σ 2. Univariate. 1 2πσ e. Kalman Filter Localization Bayes Filter Reminder Prediction Correction Gaussians p(x) ~ N(µ,σ 2 ) : Properties of Gaussians Univariate p(x) = 1 1 2πσ e 2 (x µ) 2 σ 2 µ Univariate -σ σ Multivariate µ Multivariate

More information

Estimating Time-Series Models

Estimating Time-Series Models Estimating ime-series Models he Box-Jenkins methodology for tting a model to a scalar time series fx t g consists of ve stes:. Decide on the order of di erencing d that is needed to roduce a stationary

More information

Chapter 7: Special Distributions

Chapter 7: Special Distributions This chater first resents some imortant distributions, and then develos the largesamle distribution theory which is crucial in estimation and statistical inference Discrete distributions The Bernoulli

More information

Convex Optimization methods for Computing Channel Capacity

Convex Optimization methods for Computing Channel Capacity Convex Otimization methods for Comuting Channel Caacity Abhishek Sinha Laboratory for Information and Decision Systems (LIDS), MIT sinhaa@mit.edu May 15, 2014 We consider a classical comutational roblem

More information

PROFIT MAXIMIZATION. π = p y Σ n i=1 w i x i (2)

PROFIT MAXIMIZATION. π = p y Σ n i=1 w i x i (2) PROFIT MAXIMIZATION DEFINITION OF A NEOCLASSICAL FIRM A neoclassical firm is an organization that controls the transformation of inuts (resources it owns or urchases into oututs or roducts (valued roducts

More information

Numerical Linear Algebra

Numerical Linear Algebra Numerical Linear Algebra Numerous alications in statistics, articularly in the fitting of linear models. Notation and conventions: Elements of a matrix A are denoted by a ij, where i indexes the rows and

More information

Autonomous Mobile Robot Design

Autonomous Mobile Robot Design Autonomous Mobile Robot Design Topic: Extended Kalman Filter Dr. Kostas Alexis (CSE) These slides relied on the lectures from C. Stachniss, J. Sturm and the book Probabilistic Robotics from Thurn et al.

More information

EKF, UKF. Pieter Abbeel UC Berkeley EECS. Many slides adapted from Thrun, Burgard and Fox, Probabilistic Robotics

EKF, UKF. Pieter Abbeel UC Berkeley EECS. Many slides adapted from Thrun, Burgard and Fox, Probabilistic Robotics EKF, UKF Pieter Abbeel UC Berkeley EECS Many slides adapted from Thrun, Burgard and Fox, Probabilistic Robotics Kalman Filter Kalman Filter = special case of a Bayes filter with dynamics model and sensory

More information

EKF, UKF. Pieter Abbeel UC Berkeley EECS. Many slides adapted from Thrun, Burgard and Fox, Probabilistic Robotics

EKF, UKF. Pieter Abbeel UC Berkeley EECS. Many slides adapted from Thrun, Burgard and Fox, Probabilistic Robotics EKF, UKF Pieter Abbeel UC Berkeley EECS Many slides adapted from Thrun, Burgard and Fox, Probabilistic Robotics Kalman Filter Kalman Filter = special case of a Bayes filter with dynamics model and sensory

More information

Vlad Estivill-Castro. Robots for People --- A project for intelligent integrated systems

Vlad Estivill-Castro. Robots for People --- A project for intelligent integrated systems 1 Vlad Estivill-Castro Robots for People --- A project for intelligent integrated systems V. Estivill-Castro 2 Probabilistic Map-based Localization (Kalman Filter) Chapter 5 (textbook) Based on textbook

More information

HENSEL S LEMMA KEITH CONRAD

HENSEL S LEMMA KEITH CONRAD HENSEL S LEMMA KEITH CONRAD 1. Introduction In the -adic integers, congruences are aroximations: for a and b in Z, a b mod n is the same as a b 1/ n. Turning information modulo one ower of into similar

More information

From Bayes to Extended Kalman Filter

From Bayes to Extended Kalman Filter From Bayes to Extended Kalman Filter Michal Reinštein Czech Technical University in Prague Faculty of Electrical Engineering, Department of Cybernetics Center for Machine Perception http://cmp.felk.cvut.cz/

More information

CSE 599d - Quantum Computing When Quantum Computers Fall Apart

CSE 599d - Quantum Computing When Quantum Computers Fall Apart CSE 599d - Quantum Comuting When Quantum Comuters Fall Aart Dave Bacon Deartment of Comuter Science & Engineering, University of Washington In this lecture we are going to begin discussing what haens to

More information

REFINED STRAIN ENERGY OF THE SHELL

REFINED STRAIN ENERGY OF THE SHELL REFINED STRAIN ENERGY OF THE SHELL Ryszard A. Walentyński Deartment of Building Structures Theory, Silesian University of Technology, Gliwice, PL44-11, Poland ABSTRACT The aer rovides information on evaluation

More information

Session 5: Review of Classical Astrodynamics

Session 5: Review of Classical Astrodynamics Session 5: Review of Classical Astrodynamics In revious lectures we described in detail the rocess to find the otimal secific imulse for a articular situation. Among the mission requirements that serve

More information

EE 508 Lecture 13. Statistical Characterization of Filter Characteristics

EE 508 Lecture 13. Statistical Characterization of Filter Characteristics EE 508 Lecture 3 Statistical Characterization of Filter Characteristics Comonents used to build filters are not recisely redictable R L C Temerature Variations Manufacturing Variations Aging Model variations

More information

Paper C Exact Volume Balance Versus Exact Mass Balance in Compositional Reservoir Simulation

Paper C Exact Volume Balance Versus Exact Mass Balance in Compositional Reservoir Simulation Paer C Exact Volume Balance Versus Exact Mass Balance in Comositional Reservoir Simulation Submitted to Comutational Geosciences, December 2005. Exact Volume Balance Versus Exact Mass Balance in Comositional

More information

the robot in its current estimated position and orientation (also include a point at the reference point of the robot)

the robot in its current estimated position and orientation (also include a point at the reference point of the robot) CSCI 4190 Introduction to Robotic Algorithms, Spring 006 Assignment : out February 13, due February 3 and March Localization and the extended Kalman filter In this assignment, you will write a program

More information

Extension of Minimax to Infinite Matrices

Extension of Minimax to Infinite Matrices Extension of Minimax to Infinite Matrices Chris Calabro June 21, 2004 Abstract Von Neumann s minimax theorem is tyically alied to a finite ayoff matrix A R m n. Here we show that (i) if m, n are both inite,

More information

ENGR352 Problem Set 02

ENGR352 Problem Set 02 engr352/engr352p02 September 13, 2018) ENGR352 Problem Set 02 Transfer function of an estimator 1. Using Eq. (1.1.4-27) from the text, find the correct value of r ss (the result given in the text is incorrect).

More information

Journal of Chemical and Pharmaceutical Research, 2014, 6(5): Research Article

Journal of Chemical and Pharmaceutical Research, 2014, 6(5): Research Article Available online www.jocr.com Journal of Chemical and harmaceutical Research, 04, 6(5):904-909 Research Article ISSN : 0975-7384 CODEN(USA) : JCRC5 Robot soccer match location rediction and the alied research

More information

Using the Kalman Filter to Estimate the State of a Maneuvering Aircraft

Using the Kalman Filter to Estimate the State of a Maneuvering Aircraft 1 Using the Kalman Filter to Estimate the State of a Maneuvering Aircraft K. Meier and A. Desai Abstract Using sensors that only measure the bearing angle and range of an aircraft, a Kalman filter is implemented

More information

How to calculate the current-current correlation function, correct to leading order of nonlinearity

How to calculate the current-current correlation function, correct to leading order of nonlinearity How to calculate the current-current correlation function, correct to leading order of nonlinearity Jian-Sheng Wang Deartment of Physics and Center for Comutational Science and Engineering, ational University

More information

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK

Towards understanding the Lorenz curve using the Uniform distribution. Chris J. Stephens. Newcastle City Council, Newcastle upon Tyne, UK Towards understanding the Lorenz curve using the Uniform distribution Chris J. Stehens Newcastle City Council, Newcastle uon Tyne, UK (For the Gini-Lorenz Conference, University of Siena, Italy, May 2005)

More information

arxiv: v1 [physics.data-an] 26 Oct 2012

arxiv: v1 [physics.data-an] 26 Oct 2012 Constraints on Yield Parameters in Extended Maximum Likelihood Fits Till Moritz Karbach a, Maximilian Schlu b a TU Dortmund, Germany, moritz.karbach@cern.ch b TU Dortmund, Germany, maximilian.schlu@cern.ch

More information

Einführung in Stochastische Prozesse und Zeitreihenanalyse Vorlesung, 2013S, 2.0h March 2015 Hubalek/Scherrer

Einführung in Stochastische Prozesse und Zeitreihenanalyse Vorlesung, 2013S, 2.0h March 2015 Hubalek/Scherrer Name: Mat.Nr.: Studium: Bitte keinen Rotstift verwenden! 15.593 Einführung in Stochastische Prozesse und Zeitreihenanalyse Vorlesung, 213S, 2.h March 215 Hubalek/Scherrer (Dauer 9 Minutes, Permissible

More information

Chapter 7 Rational and Irrational Numbers

Chapter 7 Rational and Irrational Numbers Chater 7 Rational and Irrational Numbers In this chater we first review the real line model for numbers, as discussed in Chater 2 of seventh grade, by recalling how the integers and then the rational numbers

More information

Statics and dynamics: some elementary concepts

Statics and dynamics: some elementary concepts 1 Statics and dynamics: some elementary concets Dynamics is the study of the movement through time of variables such as heartbeat, temerature, secies oulation, voltage, roduction, emloyment, rices and

More information

General Linear Model Introduction, Classes of Linear models and Estimation

General Linear Model Introduction, Classes of Linear models and Estimation Stat 740 General Linear Model Introduction, Classes of Linear models and Estimation An aim of scientific enquiry: To describe or to discover relationshis among events (variables) in the controlled (laboratory)

More information

Principal Components Analysis and Unsupervised Hebbian Learning

Principal Components Analysis and Unsupervised Hebbian Learning Princial Comonents Analysis and Unsuervised Hebbian Learning Robert Jacobs Deartment of Brain & Cognitive Sciences University of Rochester Rochester, NY 1467, USA August 8, 008 Reference: Much of the material

More information

Solutions to Problems in Goldstein, Classical Mechanics, Second Edition. Chapter 9

Solutions to Problems in Goldstein, Classical Mechanics, Second Edition. Chapter 9 Solutions to Problems in Goldstein, Classical Mechanics, Second Edition Homer Reid October 29, 2002 Chater 9 Problem 9. One of the attemts at combining the two sets of Hamilton s equations into one tries

More information

Kalman Filter Computer Vision (Kris Kitani) Carnegie Mellon University

Kalman Filter Computer Vision (Kris Kitani) Carnegie Mellon University Kalman Filter 16-385 Computer Vision (Kris Kitani) Carnegie Mellon University Examples up to now have been discrete (binary) random variables Kalman filtering can be seen as a special case of a temporal

More information

1 Probability Spaces and Random Variables

1 Probability Spaces and Random Variables 1 Probability Saces and Random Variables 1.1 Probability saces Ω: samle sace consisting of elementary events (or samle oints). F : the set of events P: robability 1.2 Kolmogorov s axioms Definition 1.2.1

More information

Elementary Analysis in Q p

Elementary Analysis in Q p Elementary Analysis in Q Hannah Hutter, May Szedlák, Phili Wirth November 17, 2011 This reort follows very closely the book of Svetlana Katok 1. 1 Sequences and Series In this section we will see some

More information

MODEL-BASED MULTIPLE FAULT DETECTION AND ISOLATION FOR NONLINEAR SYSTEMS

MODEL-BASED MULTIPLE FAULT DETECTION AND ISOLATION FOR NONLINEAR SYSTEMS MODEL-BASED MULIPLE FAUL DEECION AND ISOLAION FOR NONLINEAR SYSEMS Ivan Castillo, and homas F. Edgar he University of exas at Austin Austin, X 78712 David Hill Chemstations Houston, X 77009 Abstract A

More information

Quantitative estimates of propagation of chaos for stochastic systems with W 1, kernels

Quantitative estimates of propagation of chaos for stochastic systems with W 1, kernels oname manuscrit o. will be inserted by the editor) Quantitative estimates of roagation of chaos for stochastic systems with W, kernels Pierre-Emmanuel Jabin Zhenfu Wang Received: date / Acceted: date Abstract

More information

ESE 524 Detection and Estimation Theory

ESE 524 Detection and Estimation Theory ESE 524 Detection and Estimation heory Joseh A. O Sullivan Samuel C. Sachs Professor Electronic Systems and Signals Research Laboratory Electrical and Systems Engineering Washington University 2 Urbauer

More information

An Investigation on the Numerical Ill-conditioning of Hybrid State Estimators

An Investigation on the Numerical Ill-conditioning of Hybrid State Estimators An Investigation on the Numerical Ill-conditioning of Hybrid State Estimators S. K. Mallik, Student Member, IEEE, S. Chakrabarti, Senior Member, IEEE, S. N. Singh, Senior Member, IEEE Deartment of Electrical

More information

TECHNICAL RESEARCH REPORT

TECHNICAL RESEARCH REPORT TECHNICAL RESEARCH REPORT A Summary of Satellite Orbit Related Calculations by A.H. Murad, K.D. Jang, G. Atallah, R. Karne, J. Baras CSHCN T.R. 95-12 (ISR T.R. 95-107) The Center for Satellite and Hybrid

More information

Lecture 6. 2 Recurrence/transience, harmonic functions and martingales

Lecture 6. 2 Recurrence/transience, harmonic functions and martingales Lecture 6 Classification of states We have shown that all states of an irreducible countable state Markov chain must of the same tye. This gives rise to the following classification. Definition. [Classification

More information

We collect some results that might be covered in a first course in algebraic number theory.

We collect some results that might be covered in a first course in algebraic number theory. 1 Aendices We collect some results that might be covered in a first course in algebraic number theory. A. uadratic Recirocity Via Gauss Sums A1. Introduction In this aendix, is an odd rime unless otherwise

More information

Magnetospheric Physics - Homework, 4/04/2014

Magnetospheric Physics - Homework, 4/04/2014 Magnetosheric hysics - Homework, // 7. Fast wave, fast shock, erendicular shock, entroy, jum relation. Consider the fast erendicular shock. a) Determine the ositive root of the solution for the comression

More information

CHAPTER 5 TANGENT VECTORS

CHAPTER 5 TANGENT VECTORS CHAPTER 5 TANGENT VECTORS In R n tangent vectors can be viewed from two ersectives (1) they cature the infinitesimal movement along a ath, the direction, and () they oerate on functions by directional

More information

NUMERICAL ANALYSIS OF THE IMPACT OF THE INLET AND OUTLET JETS FOR THE THERMAL STRATIFICATION INSIDE A STORAGE TANK

NUMERICAL ANALYSIS OF THE IMPACT OF THE INLET AND OUTLET JETS FOR THE THERMAL STRATIFICATION INSIDE A STORAGE TANK NUMERICAL ANALYSIS OF HE IMAC OF HE INLE AND OULE JES FOR HE HERMAL SRAIFICAION INSIDE A SORAGE ANK A. Zachár I. Farkas F. Szlivka Deartment of Comuter Science Szent IstvÆn University Æter K. u.. G d llı

More information

Automatic Generation and Integration of Equations of Motion for Linked Mechanical Systems

Automatic Generation and Integration of Equations of Motion for Linked Mechanical Systems Automatic Generation and Integration of Equations of Motion for Linked Mechanical Systems D. Todd Griffith a, John L. Junkins a, and James D. Turner b a Deartment of Aerosace Engineering, Texas A&M University,

More information

Elements of Asymptotic Theory. James L. Powell Department of Economics University of California, Berkeley

Elements of Asymptotic Theory. James L. Powell Department of Economics University of California, Berkeley Elements of Asymtotic Theory James L. Powell Deartment of Economics University of California, Berkeley Objectives of Asymtotic Theory While exact results are available for, say, the distribution of the

More information

An Inverse Problem for Two Spectra of Complex Finite Jacobi Matrices

An Inverse Problem for Two Spectra of Complex Finite Jacobi Matrices Coyright 202 Tech Science Press CMES, vol.86, no.4,.30-39, 202 An Inverse Problem for Two Sectra of Comlex Finite Jacobi Matrices Gusein Sh. Guseinov Abstract: This aer deals with the inverse sectral roblem

More information

The Quark-Parton Model

The Quark-Parton Model The Quark-Parton Model Before uarks and gluons were generally acceted Feynman roosed that the roton was made u of oint-like constituents artons Both Bjorken Scaling and the Callan-Gross relationshi can

More information

Autonomous Navigation for Flying Robots

Autonomous Navigation for Flying Robots Computer Vision Group Prof. Daniel Cremers Autonomous Navigation for Flying Robots Lecture 6.2: Kalman Filter Jürgen Sturm Technische Universität München Motivation Bayes filter is a useful tool for state

More information

Higher order theory for analytic saddle point approximations to the Ρ Ρ and Ρ Ś reflected arrivals at a solid/solid interface

Higher order theory for analytic saddle point approximations to the Ρ Ρ and Ρ Ś reflected arrivals at a solid/solid interface Higher order theory for analytic saddle oint aroximations to the Ρ Ρ and Ρ Ś reflected arrivals at a solid/solid interface P.F Daley ABSTACT The high frequency solution to the roblem of a Ρ P and Ρ S reflected

More information

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL

MODELING THE RELIABILITY OF C4ISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Technical Sciences and Alied Mathematics MODELING THE RELIABILITY OF CISR SYSTEMS HARDWARE/SOFTWARE COMPONENTS USING AN IMPROVED MARKOV MODEL Cezar VASILESCU Regional Deartment of Defense Resources Management

More information

LPC methods are the most widely used in. recognition, speaker recognition and verification

LPC methods are the most widely used in. recognition, speaker recognition and verification Digital Seech Processing Lecture 3 Linear Predictive Coding (LPC)- Introduction LPC Methods LPC methods are the most widely used in seech coding, seech synthesis, seech recognition, seaker recognition

More information

University of North Carolina-Charlotte Department of Electrical and Computer Engineering ECGR 4143/5195 Electrical Machinery Fall 2009

University of North Carolina-Charlotte Department of Electrical and Computer Engineering ECGR 4143/5195 Electrical Machinery Fall 2009 University of North Carolina-Charlotte Deartment of Electrical and Comuter Engineering ECG 4143/5195 Electrical Machinery Fall 9 Problem Set 5 Part Due: Friday October 3 Problem 3: Modeling the exerimental

More information

Robot Localization and Kalman Filters

Robot Localization and Kalman Filters Robot Localization and Kalman Filters Rudy Negenborn rudy@negenborn.net August 26, 2003 Outline Robot Localization Probabilistic Localization Kalman Filters Kalman Localization Kalman Localization with

More information

Improved Bounds on Bell Numbers and on Moments of Sums of Random Variables

Improved Bounds on Bell Numbers and on Moments of Sums of Random Variables Imroved Bounds on Bell Numbers and on Moments of Sums of Random Variables Daniel Berend Tamir Tassa Abstract We rovide bounds for moments of sums of sequences of indeendent random variables. Concentrating

More information

Feedback-error control

Feedback-error control Chater 4 Feedback-error control 4.1 Introduction This chater exlains the feedback-error (FBE) control scheme originally described by Kawato [, 87, 8]. FBE is a widely used neural network based controller

More information

The Arm Prime Factors Decomposition

The Arm Prime Factors Decomposition The Arm Prime Factors Decomosition Arm Boris Nima arm.boris@gmail.com Abstract We introduce the Arm rime factors decomosition which is the equivalent of the Taylor formula for decomosition of integers

More information

ON MINKOWSKI MEASURABILITY

ON MINKOWSKI MEASURABILITY ON MINKOWSKI MEASURABILITY F. MENDIVIL AND J. C. SAUNDERS DEPARTMENT OF MATHEMATICS AND STATISTICS ACADIA UNIVERSITY WOLFVILLE, NS CANADA B4P 2R6 Abstract. Two athological roerties of Minkowski content

More information

Radial Basis Function Networks: Algorithms

Radial Basis Function Networks: Algorithms Radial Basis Function Networks: Algorithms Introduction to Neural Networks : Lecture 13 John A. Bullinaria, 2004 1. The RBF Maing 2. The RBF Network Architecture 3. Comutational Power of RBF Networks 4.

More information

Elements of Asymptotic Theory. James L. Powell Department of Economics University of California, Berkeley

Elements of Asymptotic Theory. James L. Powell Department of Economics University of California, Berkeley Elements of Asymtotic Theory James L. Powell Deartment of Economics University of California, Berkeley Objectives of Asymtotic Theory While exact results are available for, say, the distribution of the

More information

4. Score normalization technical details We now discuss the technical details of the score normalization method.

4. Score normalization technical details We now discuss the technical details of the score normalization method. SMT SCORING SYSTEM This document describes the scoring system for the Stanford Math Tournament We begin by giving an overview of the changes to scoring and a non-technical descrition of the scoring rules

More information

F(p) y + 3y + 2y = δ(t a) y(0) = 0 and y (0) = 0.

F(p) y + 3y + 2y = δ(t a) y(0) = 0 and y (0) = 0. Page 5- Chater 5: Lalace Transforms The Lalace Transform is a useful tool that is used to solve many mathematical and alied roblems. In articular, the Lalace transform is a technique that can be used to

More information

Using Factor Analysis to Study the Effecting Factor on Traffic Accidents

Using Factor Analysis to Study the Effecting Factor on Traffic Accidents Using Factor Analysis to Study the Effecting Factor on Traffic Accidents Abstract Layla A. Ahmed Deartment of Mathematics, College of Education, University of Garmian, Kurdistan Region Iraq This aer is

More information

7A Midterm II - Lin Problem 4 Solutions

7A Midterm II - Lin Problem 4 Solutions Œ Š? U Œ Œ $ ` Œ l U d Œ d l m J!#"$%'&)(*+&-,/.0132 452768+'9:.;,/

More information

Introduction to Unscented Kalman Filter

Introduction to Unscented Kalman Filter Introduction to Unscented Kalman Filter 1 Introdution In many scientific fields, we use certain models to describe the dynamics of system, such as mobile robot, vision tracking and so on. The word dynamics

More information

Explicit Lp-norm estimates of infinitely divisible random vectors in Hilbert spaces with applications

Explicit Lp-norm estimates of infinitely divisible random vectors in Hilbert spaces with applications University of Tennessee, Knoxville Trace: Tennessee Research and Creative Exchange Doctoral Dissertations Graduate School 5-211 Exlicit L-norm estimates of infinitely divisible random vectors in Hilbert

More information

A Study of Covariances within Basic and Extended Kalman Filters

A Study of Covariances within Basic and Extended Kalman Filters A Study of Covariances within Basic and Extended Kalman Filters David Wheeler Kyle Ingersoll December 2, 2013 Abstract This paper explores the role of covariance in the context of Kalman filters. The underlying

More information

Image Alignment and Mosaicing Feature Tracking and the Kalman Filter

Image Alignment and Mosaicing Feature Tracking and the Kalman Filter Image Alignment and Mosaicing Feature Tracking and the Kalman Filter Image Alignment Applications Local alignment: Tracking Stereo Global alignment: Camera jitter elimination Image enhancement Panoramic

More information

Participation Factors. However, it does not give the influence of each state on the mode.

Participation Factors. However, it does not give the influence of each state on the mode. Particiation Factors he mode shae, as indicated by the right eigenvector, gives the relative hase of each state in a articular mode. However, it does not give the influence of each state on the mode. We

More information

Applied Fitting Theory VI. Formulas for Kinematic Fitting

Applied Fitting Theory VI. Formulas for Kinematic Fitting Alied Fitting heory VI Paul Avery CBX 98 37 June 9, 1998 Ar. 17, 1999 (rev.) I Introduction Formulas for Kinematic Fitting I intend for this note and the one following it to serve as mathematical references,

More information

I Poles & zeros. I First-order systems. I Second-order systems. I E ect of additional poles. I E ect of zeros. I E ect of nonlinearities

I Poles & zeros. I First-order systems. I Second-order systems. I E ect of additional poles. I E ect of zeros. I E ect of nonlinearities EE C28 / ME C34 Lecture Chater 4 Time Resonse Alexandre Bayen Deartment of Electrical Engineering & Comuter Science University of California Berkeley Lecture abstract Toics covered in this resentation

More information

Distributed Rule-Based Inference in the Presence of Redundant Information

Distributed Rule-Based Inference in the Presence of Redundant Information istribution Statement : roved for ublic release; distribution is unlimited. istributed Rule-ased Inference in the Presence of Redundant Information June 8, 004 William J. Farrell III Lockheed Martin dvanced

More information

15-451/651: Design & Analysis of Algorithms October 23, 2018 Lecture #17: Prediction from Expert Advice last changed: October 25, 2018

15-451/651: Design & Analysis of Algorithms October 23, 2018 Lecture #17: Prediction from Expert Advice last changed: October 25, 2018 5-45/65: Design & Analysis of Algorithms October 23, 208 Lecture #7: Prediction from Exert Advice last changed: October 25, 208 Prediction with Exert Advice Today we ll study the roblem of making redictions

More information

CALIFORNIA INSTITUTE OF TECHNOLOGY Control and Dynamical Systems. CDS 110b

CALIFORNIA INSTITUTE OF TECHNOLOGY Control and Dynamical Systems. CDS 110b CALIFORNIA INSTITUTE OF TECHNOLOGY Control and Dynamical Systems CDS 110b R. M. Murray Kalman Filters 25 January 2006 Reading: This set of lectures provides a brief introduction to Kalman filtering, following

More information

CS 532: 3D Computer Vision 6 th Set of Notes

CS 532: 3D Computer Vision 6 th Set of Notes 1 CS 532: 3D Computer Vision 6 th Set of Notes Instructor: Philippos Mordohai Webpage: www.cs.stevens.edu/~mordohai E-mail: Philippos.Mordohai@stevens.edu Office: Lieb 215 Lecture Outline Intro to Covariance

More information

A Simple Weight Decay Can Improve. Abstract. It has been observed in numerical simulations that a weight decay can improve

A Simple Weight Decay Can Improve. Abstract. It has been observed in numerical simulations that a weight decay can improve In Advances in Neural Information Processing Systems 4, J.E. Moody, S.J. Hanson and R.P. Limann, eds. Morgan Kaumann Publishers, San Mateo CA, 1995,. 950{957. A Simle Weight Decay Can Imrove Generalization

More information

Chapter 6. Phillip Hall - Room 537, Huxley

Chapter 6. Phillip Hall - Room 537, Huxley Chater 6 6 Partial Derivatives.................................................... 72 6. Higher order artial derivatives...................................... 73 6.2 Matrix of artial derivatives.........................................74

More information

PArtially observable Markov decision processes

PArtially observable Markov decision processes Solving Continuous-State POMDPs via Density Projection Enlu Zhou, Member, IEEE, Michael C. Fu, Fellow, IEEE, and Steven I. Marcus, Fellow, IEEE Abstract Research on numerical solution methods for artially

More information

Lecture 8: Signal Detection and Noise Assumption

Lecture 8: Signal Detection and Noise Assumption ECE 830 Fall 0 Statistical Signal Processing instructor: R. Nowak Lecture 8: Signal Detection and Noise Assumption Signal Detection : X = W H : X = S + W where W N(0, σ I n n and S = [s, s,..., s n ] T

More information

arxiv: v1 [cs.ro] 24 May 2017

arxiv: v1 [cs.ro] 24 May 2017 A Near-Otimal Searation Princile for Nonlinear Stochastic Systems Arising in Robotic Path Planning and Control Mohammadhussein Rafieisakhaei 1, Suman Chakravorty 2 and P. R. Kumar 1 arxiv:1705.08566v1

More information

Chapter 3. GMM: Selected Topics

Chapter 3. GMM: Selected Topics Chater 3. GMM: Selected oics Contents Otimal Instruments. he issue of interest..............................2 Otimal Instruments under the i:i:d: assumtion..............2. he basic result............................2.2

More information

Residual Stress and Deformation Modelling for Metal Additive Manufacturing Processes

Residual Stress and Deformation Modelling for Metal Additive Manufacturing Processes Proceedings of the World Congress on Mechanical, Chemical, and Material Engineering (MCM 215) Barcelona, Sain July 2-21, 215 Paer No. 245 Residual Stress and Deformation Modelling for Metal Additive Manufacturing

More information

SECTION 5: FIBRATIONS AND HOMOTOPY FIBERS

SECTION 5: FIBRATIONS AND HOMOTOPY FIBERS SECTION 5: FIBRATIONS AND HOMOTOPY FIBERS In this section we will introduce two imortant classes of mas of saces, namely the Hurewicz fibrations and the more general Serre fibrations, which are both obtained

More information

Series Handout A. 1. Determine which of the following sums are geometric. If the sum is geometric, express the sum in closed form.

Series Handout A. 1. Determine which of the following sums are geometric. If the sum is geometric, express the sum in closed form. Series Handout A. Determine which of the following sums are geometric. If the sum is geometric, exress the sum in closed form. 70 a) k= ( k ) b) 50 k= ( k )2 c) 60 k= ( k )k d) 60 k= (.0)k/3 2. Find the

More information

9 The Theory of Special Relativity

9 The Theory of Special Relativity 9 The Theory of Secial Relativity Assign: Read Chater 4 of Carrol and Ostlie (2006) Newtonian hysics is a quantitative descrition of Nature excet under three circumstances: 1. In the realm of the very

More information

On split sample and randomized confidence intervals for binomial proportions

On split sample and randomized confidence intervals for binomial proportions On slit samle and randomized confidence intervals for binomial roortions Måns Thulin Deartment of Mathematics, Usala University arxiv:1402.6536v1 [stat.me] 26 Feb 2014 Abstract Slit samle methods have

More information

Availability and Maintainability. Piero Baraldi

Availability and Maintainability. Piero Baraldi Availability and Maintainability 1 Introduction: reliability and availability System tyes Non maintained systems: they cannot be reaired after a failure (a telecommunication satellite, a F1 engine, a vessel

More information

Principles of Computed Tomography (CT)

Principles of Computed Tomography (CT) Page 298 Princiles of Comuted Tomograhy (CT) The theoretical foundation of CT dates back to Johann Radon, a mathematician from Vienna who derived a method in 1907 for rojecting a 2-D object along arallel

More information

Linear diophantine equations for discrete tomography

Linear diophantine equations for discrete tomography Journal of X-Ray Science and Technology 10 001 59 66 59 IOS Press Linear diohantine euations for discrete tomograhy Yangbo Ye a,gewang b and Jiehua Zhu a a Deartment of Mathematics, The University of Iowa,

More information

Positivity, local smoothing and Harnack inequalities for very fast diffusion equations

Positivity, local smoothing and Harnack inequalities for very fast diffusion equations Positivity, local smoothing and Harnack inequalities for very fast diffusion equations Dedicated to Luis Caffarelli for his ucoming 60 th birthday Matteo Bonforte a, b and Juan Luis Vázquez a, c Abstract

More information

CFD AS A DESIGN TOOL FOR FLUID POWER COMPONENTS

CFD AS A DESIGN TOOL FOR FLUID POWER COMPONENTS CFD AS A DESIGN TOOL FOR FLUID POWER COMPONENTS M. BORGHI - M. MILANI Diartimento di Scienze dell Ingegneria Università degli Studi di Modena Via Cami, 213/b 41100 Modena E-mail: borghi@omero.dsi.unimo.it

More information