Kalman Filters for Mapping and Localization
|
|
- Clifton Fox
- 6 years ago
- Views:
Transcription
1 Kalman Filters for Mapping and Localization
2
3
4
5 Sensors If you can t model the world, then sensors are the robot s link to the external world (obsession with depth) Laser Kinect IR rangefinder sonar rangefinder Light-field camera
6 Sensors Robots link to the external world... Sensors, sensors, sensors! and tracking what is sensed: world models Gyroscope (orientatiation) GPS Force/ Torque Inertial measurement unit (gyro + accelerometer) Compass
7 Infrared sensors Noncontact bump sensor (1) sensing is based on light intensity. object-sensing IR looks for changes at this distance diffuse distance-sensing IR (2) sensing is based on angle receved. IR emitter/detector pair , Howie Choset with slides from and Z. Dodds
8 Infrared Calibration The response to white copy paper (a dull, reflective surface) raw values (put into 4 bits) inches 15º increments in the dark , Howie Choset with slides from and Z. Dodds fluorescent light incandescent light
9 Infrared Calibration energy vs. distance for various materials ( the incident angle is 0º, or head-on ) ( with no ambient light ) , Howie Choset with slides from and Z. Dodds
10 Sonar sensing single-transducer sonar timeline 0 a chirp is emitted into the environment 75µs typically when reverberations from the initial chirp have stopped the transducer goes into receiving mode and awaits a signal... limiting range sensing.5s after a short time, the signal will be too weak to be detected , Howie Choset with slides from and Z. Dodds blanking time Polaroid sonar emitter/receivers No lower range limit for paired sonars...
11 Sonar effects (a) Sonar providing an accurate range measurement (b-c) Lateral resolution is not very precise; the closest object in the beam s cone provides the response (d) Specular reflections cause walls to disappear (e) Open corners produce a weak spherical wavefront (f) Closed corners measure to the corner itself because of multiple reflections --> sonar ray tracing , Howie Choset with slides from and Z. Dodds
12 Sonar modeling initial time response accumulated responses blanking time cone width spatial response , Howie Choset with slides from and Z. Dodds
13 Laser Ranging LIDAR/Laser range finder , Howie Choset with slides from and Z. Dodds LIDAR map
14 Souped-Up Robots
15 The 3d time-of-flight camera Recent, Cool... (1) RF-modulated optical radiation field output (2) Reflects off the environment (3) Time-interval input measurements provides data (not time-integrated!) How can we get an image from this information?
16 More recent, Cooler... Structured light: Project a known dot pattern with an IR transmitter (invisible to humans) Infer depth from deformation to that pattern depth from focus: Points far away are blurry depth from stereo: Closer points are shifted
17 The Latest, Coolest... Light field camera (passive) Capture the light going in every direction at every 3D point ieee.org
18 The Problem Mapping: What is the world around me (geometry, landmarks) sense from various positions integrate measurements to produce map assumes perfect knowledge of position Localization: Where am I in the world (position wrt landmarks) sense relate sensor readings to a world model compute location relative to model assumes a perfect world model Together, these are SLAM (Simultaneous Localization and Mapping) How can you localize without a map? How can you map without localization? All localization, mapping or SLAM methods are based on updating a state: What makes a state? Localization? Map? Both? How certain is the state?
19 Representations for Bayesian Robot Localization Discrete approaches ( 95) Topological representation ( 95) uncertainty handling (POMDPs) occas. global localization, recovery Grid-based, metric representation ( 96) global localization, recovery Particle filters ( 99) sample-based representation global localization, recovery AI Kalman filters (late-80s?) Gaussians approximately linear models position tracking Robotics Multi-hypothesis ( 00) multiple Kalman filters global localization, recovery
20 Gaussian (or Normal) Distribution Univariate pp xx ~NN μμ, σσ 2 pp xx = 1 2ππσσ ee 1(xx μμ) 2 2 σσ 2 µ -σ σ Multivariate pp xx ~NN μμ, Σ pp xx = 1 (2ππ) dd/2 Σ 1 ee 2 xx μμ TT Σ 1 (xx μμ) 1/2
21 Properties of Gaussians XX~NN μμ, σσ 2 YY = aaaa + bb YY~NN(aaaa + bb, aa2 σσ 2 ) XX 1 ~NN μμ 1, σσ XX 2 ~NN μμ 2, σσ XX 1 + XX 2 ~NN(μμ 1 + μμ 2, σσ 1 + σσ 2 ) We stay in the Gaussian world as long as we start with Gaussians and perform only linear transformations. Same holds for multivariate Gaussians
22 Dynamical Systems System that changes We will represent a system by its state xx tt at time t We will also require that a system be observable Jeb Corliss Oliver uu tt uu tt+1 uu tt+1 external command input xx tt xx tt+1 xx tt+2 zz tt zz tt+1 zz tt+2 process model (state transition) observation model (measurements of the states)
23 Kalman Filter Seminal paper published in 1960 Great web page at Recursive solution for discrete linear filtering problems A state x R n A measurement z R m Discrete (i.e. for time t = 1, 2, 3, ) Recursive process (i.e. x t = f ( x t-1 ) ) Linear system (i.e x t = A x t-1 ) The system is defined by: 1) linear process model 2) linear measurement model x t = A x t-1 + B u t-1 + w t-1 state control transition Input (optional) Gaussian white noise How a state transitions into another state z t = H x t + v t observation model Gaussian white noise How a state relates to a measurement
24 Kalman Filter 1. Prior estimate xx tt at step t : Use the process model to predict what will be the next state of the robot 2. Posterior estimate xx tt at step t : Use the observation model to correct the prediction by using sensor measurement Compute posterior estimate as a linear combination of the prior estimate and difference between the actual measurement and expected measurement x ˆ = xˆ + K ( z Hxˆ ) t t t Where the Kalman gain KK tt is a blending factor that adds a measurement innovation. t t
25 Kalman Filter Define prior error between true state and prior estimate and the prior covariance as Define posterior error between true state and posterior estimate and it s posterior covariance as The gain K that minimizes the posterior covariance is defined by Note that K t e = x xˆ t Σ t = E( e te t e t If RR 0 then KK tt = HH 1 and xx tt = xx tt + KK tt zz tt HHxx tt = xx If Σ tt 0 then KK tt = 0 and xx tt = xx tt = x t t xˆ Σt = E( ete t T t T t T T = Σ H ( HΣ H + R) t t ) ) 1
26 Kalman Filter Recipe: 1. Start with an initial guess xx 0, Σ 0 2. Compute prior (prediction) xx tt = AAxx tt 1 + BBuu tt 1 Σ tt = AAΣ tt 1 AA TT + QQ Time update (predict) REPEAT 3. Compute posterior (correction) KK tt = Σ tt HH TT HHΣ tt HH TT + RR 1 xx tt = xx tt + KK tt zz tt HHxx tt Σ tt = 1 KK tt HH Σ tt Measurement update (correct)
27 Initial xx 0, Σ 0 Predict xx 1 from xx 0 and uu 0 xx 1 = AAxx 0 + BBuu 0 Σ 1 = AAΣ 0 AA TT + QQ Correction using zz 1 KK 1 = Σ 1 HH TT HHΣ 1 HH TT + RR 1 xx 1 = xx 1 + KK 1 zz 1 HHxx 1 Σ 1 = 1 KK 1 HH Σ 1 Predict xx 2 from xx 1 and uu 1 xx 2 = AAxx 1 + BBuu 1 Σ 2 = AAΣ 1 AA TT + QQ , Howie Choset with slides from and Z. Dodds
28 Observability zz tt = HHxx tt If H does not provide a one-to-one mapping between the state and the measurement, then the system is unobservable zz tt = xx tt In this case H is singular, such that many states can generate the same observation KK tt = Σ tt HH TT HHΣ tt HH TT + RR 1 rank AAAA TT = rank AA rank AA + BB rank AA + rank(bb)
29 Fully Observable vs Partially Observable , Howie Choset
30 Kalman Filter Limitations Assumptions: Linear process model xx tt+1 = AAxx tt + BBuu tt + ww tt Linear observation model zz tt = HHxx tt + vv tt White Gaussian noise NN(0, Σ ) What can we do if system is not linear? Non-linear state dynamics xx tt+1 = ff xx tt, uu tt, ww tt Non-linear observations zz tt = h xx tt, vv tt Linearize it! xx tt+1 xx tt+1 + AA xx tt xx tt + WWww tt zz tt zz tt + HH xx tt xx tt + VVvv tt xx tt+1 = ff xx tt, uu tt, 0 zz tt = h(xx tt, 0)
31 Extended Kalman Filter Where A, H, W and V are Jacobians defined by AA xx tt = ff 1 (xx tt, uu tt, 0) ff 1(xx tt, uu tt, 0) xx 1 xx NN ff MM (xx tt, uu tt, 0) ff MM(xx tt, uu tt, 0) xx 1 xx NN WW xx tt = ff 1 (xx tt, uu tt, 0) ff 1(xx tt, uu tt, 0) ww 1 ww NN ff MM (xx tt, uu tt, 0) ff MM(xx tt, uu tt, 0) ww 1 ww NN HH xx tt = h 1 (xx tt, 0) h 1(xx tt, 0) xx 1 xx NN h MM (xx tt, 0) h MM(xx tt, 0) xx 1 xx NN VV xx tt = h 1 (xx tt, 0) h 1(xx tt, 0) vv 1 vv NN h MM (xx tt, 0) h MM(xx tt, 0) vv 1 vv NN
32 EKF for Range-Bearing Localization State ss tt = xx tt yy tt θθ tt TT 2D position and orientation Input uu tt = vv tt ωω tt TT linear and angular velocity Process model ff ss tt, uu tt, ww tt = xx tt 1 + ( tt)vv tt 1 cos (θθ tt 1 ) yy tt 1 + ( tt)vv tt 1 sin (θθ tt 1 ) θθ tt 1 + ( tt)ωω tt 1 + ww xxtt ww yy ww θθtt y y t l 1 r 1 b 1 x t l N Given a map, the robot sees N landmarks with coordinates ll 1 = xx ll1 yy ll1 TT,, ll NN = xx llnn yy llnn TT The observation model is x zz tt = hh 1 (ss tt, vv 1 ) hh NN (ss tt, vv NN ) hh ii ss tt, vv tt = xx tt xx llii 2 + yytt yy llii 2 tan 1 + yy tt yy llii θθ xx tt xx tt llii vv rr vv bb
33 Linearize Process Model ff ss tt, uu tt, ww tt = xx tt 1 + ( tt)vv tt 1 cos (θθ tt 1 ) yy tt 1 + ( tt)vv tt 1 sin (θθ tt 1 ) θθ tt 1 + ( tt)ωω tt 1 + ww xxtt ww yy ww θθtt AA xx tt = ff 1 (xx tt, uu tt, 0) ff 1(xx tt, uu tt, 0) xx 1 xx NN ff MM (xx tt, uu tt, 0) ff MM(xx tt, uu tt, 0) xx 1 xx NN AA = 1 0 tt vv tt 1 sin (θθ tt 1 ) 0 1 tt vv tt 1 cos (θθ tt 1) 0 0 1
34 Linearize Observation Model zz tt = hh 1 (ss tt, vv 1 ) hh NN (ss tt, vv NN ) hh ii ss tt, vv tt = xx tt xx llii 2 + yytt yy llii 2 tan 1 yy tt yy + llii θθ xx tt xx tt llii vv rr vv bb HH ss tt = hh 1 (ss tt, 0) xx tt hh 1 (ss tt, 0) hh 1 (ss tt, 0) yy tt θθ tt hh NN (ss tt, 0) xx tt yy tt θθ tt hh NN (ss tt, 0) hh NN (ss tt, 0)
35 Extended Kalman Filter Kalman Filter Recipe: Given Extended Kalman Filter Recipe: Given Prediction Prediction Measurement correction Measurement correction
36 Kalman Filter for Dead Reckoning Robot moves along a straight line with state x = [ p, v ] T p: position v: velocity u is the input force applied to the robot Newton s 2 nd law first order finite difference: Integrate velocity Robot has velocity sensor Integrate acceleration zz tt = 0 1 pp tt vv tt The robot position is not measured The measured velocity depends on the robot velocity (duh!)
37 Example Let plug some numbers m = 1 t = 0.5
38 From Localization to Mapping For us, the landmarks have been a known quantity (we have a map with the coordinates of the landmarks), but landmarks are not part of the state Two choices: Make the state the location of the landmarks relative to the robot (I also know exactly where I am ) - No notion of location relative to past history - No fixed reference for landmarks irobot Make the state the robot location now (relative to where we started) plus landmark locations + Landmarks now have fixed location - Knowledge of my location slowly degrades (but this is inevitable )
39 EKF for Range-Bearing Localization State ss tt = xx tt yy tt θθ tt TT 2D position and orientation Input uu tt = vv tt ωω tt TT linear and angular velocity Process model ff ss tt, uu tt, ww tt = xx tt 1 + ( tt)vv tt 1 cos (θθ tt 1 ) yy tt 1 + ( tt)vv tt 1 sin (θθ tt 1 ) θθ tt 1 + ( tt)ωω tt 1 + ww xxtt ww yy ww θθtt y y t l 1 r 1 b 1 x t l N Given a map, the robot sees N landmarks with coordinates ll 1 = xx ll1 yy ll1 TT,, ll NN = xx llnn yy llnn TT The observation model is x zz tt = hh 1 (ss tt, vv 1 ) hh NN (ss tt, vv NN ) hh ii ss tt, vv tt = xx tt xx llii 2 + yytt yy llii 2 tan 1 + yy tt yy llii θθ xx tt xx tt llii vv rr vv bb
40 Data Association From observation model, we have an expected Observation of landmark #1 Observation of landmark #N So if we have N landmarks l 1,, l N and we are given a scan z t, how do associate each landmark to a scan observation? msdn.microsoft.com ll iitt = ll iitt + KK tt zz iitt hh ii xx tt, 0 Observation of landmark #i must be in the i th row!
41 Natural Visual Landmarks Visual landmarks: SIFT, SURF, corners, Se IJRR
42 Artificial Landmarks Lippsett
43 EKF for Range-Bearing Localization State ss tt = xx tt yy tt θθ tt TT 2D position and orientation Input uu tt = vv tt ωω tt TT linear and angular velocity Process model ff ss tt, uu tt, ww tt = xx tt 1 + ( tt)vv tt 1 cos (θθ tt 1 ) yy tt 1 + ( tt)vv tt 1 sin (θθ tt 1 ) θθ tt 1 + ( tt)ωω tt 1 + ww xxtt ww yy ww θθtt y y t l 1 r 1 b 1 x t l N Given a map, the robot sees N landmarks with coordinates ll 1 = xx ll1 yy ll1 TT,, ll NN = xx llnn yy llnn TT The observation model is x zz tt = hh 1 (ss tt, vv 1 ) hh NN (ss tt, vv NN ) hh ii ss tt, vv tt = xx tt xx llii 2 + yytt yy llii 2 tan 1 + yy tt yy llii θθ xx tt xx tt llii vv rr vv bb
44 Kalman Filters and SLAM Localization: state is the location of the robot Mapping: state is the location of 2D landmarks SLAM: state combines both If the state is then we can write a linear observation system note that if we don t have some fixed landmarks, our system is unobservable (we can t fully determine all unknown quantities) Covariance Σ is represented by
45 EKF Range Bearing SLAM Prior State Estimation State ss tt = xx tt yy tt θθ tt ll 1 TT ll NN TT position/orientation of robot and landmarks coordinates Input uu tt = vv tt ωω tt TT forward and angular velocity The process model for localization is xx tt 1 ttvv tt 1 cos (θθ tt 1 ) ss tt = yy tt 1 + ttvv tt 1 sin (θθ tt 1 ) θθ tt 1 ttωω tt 1 This model is augmented for 2N+3 dimensions to accommodate landmarks. This results in the process equation xx tt yy tt θθ tt ll 1tt ll NNtt = xx tt 1 yy tt 1 θθ tt 1 ll 1tt 1 ll NNtt ttvv tt 1 cos (θθ tt 1 ) ttvv tt 1 sin (θθ tt 1 ) ttωω tt 1 Landmarks don t depend on external input
46 EKF Range Bearing SLAM Prior Covariance Update We assume static landmarks. Therefore, the function f(s,u,w) only affects the robot s location and not the landmarks. Σ tt = AA tt Σ tt 1 AA tt TT + WW tt QQWW tt TT Jacobian of the process model The motion of the robot does not affect the coordinates of the landmarks AA = xx xx θθ xx θθ θθ 0 θθ θθ θθ 0 II Jacobian of the robot motion 2Nx2N identity
47 EKF Range Bearing SLAM Kalman Gain Compute the Jacobian H i of each h i and then stack them into one big matrix H. Note that h i only depends on 5 variables: x t, y t, θ t, x li, y li Need to be in the correct columns of H
48 EKF Range Bearing SLAM Measurement Observe N landmarks zz iitt = rr iitt φφ iitt TT Must have data association Which measured landmark corresponds to h i? If s t contains the coordinates of N landmarks in the map, h i predicts the measurement of each landmark Must figure which measured landmark corresponds to h i ss tt = ss tt + KK tt zz tt hh ss tt, 0 Make sure that each landmark observation zz iitt appears in the correct rows of zz tt ll iitt = ll iitt + KK tt zz iitt hh ii ss tt, 0
49 EKF Range Bearing SLAM Posterior Update From K t and H update the posterior state estimate ss tt = ss tt + KK tt zz tt hh ss tt, 0 Σ tt = II KK tt HH tt Σ tt Tada! And we are done!
50 Bearing-Only SLAM Lets assume one landmark for now s = h( s) = tan z = h(s) wikipedia [ x y θ ] T 1 l x y l l x l y y x θ Tully 2008 Often use omni-directional sensor
51 Why Bearing-Only SLAM is Challenging We cannot estimate the landmark location with one measurement We must guess the range and initialize with a large covariance due to the lack of range information The location is very uncertain and difficult to resolve with low parallax measurements The measurement model is very nonlinear, which breaks conventional filtering techniques Tully 2008
52 Bearing-Only SLAM with EKF EKF uses the standard Kalman update The Kalman gain is computed through a linearization about the current estimate The result diverges Very dependent on the initialization guess of landmarks
53 Mono SLAM A visual landmark with a single camera does not provide range Data association is given by tracking or matching visual descriptors/patches Robot Vision, Imperial College
54 Submaps As landmarks are added, the state vector grows For large maps only a few landmarks can be visible at any given time Use submaps to reduce the number of landmarks to manageable numbers
55 Navigation: RMS Titanic Leonard & Eustice EKF-based system 866 images 3494 camera constraints Path length 3.1km 2D / 3.4km 3D Convex hull > 3100m min. data / 39 min. ESDF* *excludes image registration time
56 Search of Flight 370 Inertial Navigation System (INS): A glorified IMU with Kalman filter
57 Basic system modeling ideas Summary Kalman filter as an estimation method from a system model Linearization as a way of attacking a wider variety of problems Mapping localization and mapping into EKF Extensions for managing landmark matching and not-wellconstrained systems.
Control of Mobile Robots
Control of Mobile Robots Regulation and trajectory tracking Prof. Luca Bascetta (luca.bascetta@polimi.it) Politecnico di Milano Dipartimento di Elettronica, Informazione e Bioingegneria Organization and
More informationBayes 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 informationIntroduction to Mobile Robotics Probabilistic Sensor Models
Introduction to Mobile Robotics Probabilistic Sensor Models Wolfram Burgard 1 Sensors for Mobile Robots Contact sensors: Bumpers Proprioceptive sensors Accelerometers (spring-mounted masses) Gyroscopes
More informationVariations. ECE 6540, Lecture 02 Multivariate Random Variables & Linear Algebra
Variations ECE 6540, Lecture 02 Multivariate Random Variables & Linear Algebra Last Time Probability Density Functions Normal Distribution Expectation / Expectation of a function Independence Uncorrelated
More informationECE 6540, Lecture 06 Sufficient Statistics & Complete Statistics Variations
ECE 6540, Lecture 06 Sufficient Statistics & Complete Statistics Variations Last Time Minimum Variance Unbiased Estimators Sufficient Statistics Proving t = T(x) is sufficient Neyman-Fischer Factorization
More informationObject Tracking using Kalman and Particle filtering Techniques
Object Tracking using Kalman and Particle filtering Techniques Department of Electronics and Communication Engineering National Institute of Technology Rourkela Rourkela, Odisha, 769008, India 2015 Object
More informationModule 7 (Lecture 27) RETAINING WALLS
Module 7 (Lecture 27) RETAINING WALLS Topics 1.1 RETAINING WALLS WITH METALLIC STRIP REINFORCEMENT Calculation of Active Horizontal and vertical Pressure Tie Force Factor of Safety Against Tie Failure
More informationL11. EKF SLAM: PART I. NA568 Mobile Robotics: Methods & Algorithms
L11. EKF SLAM: PART I NA568 Mobile Robotics: Methods & Algorithms Today s Topic EKF Feature-Based SLAM State Representation Process / Observation Models Landmark Initialization Robot-Landmark Correlation
More informationSupport Vector Machines. CSE 4309 Machine Learning Vassilis Athitsos Computer Science and Engineering Department University of Texas at Arlington
Support Vector Machines CSE 4309 Machine Learning Vassilis Athitsos Computer Science and Engineering Department University of Texas at Arlington 1 A Linearly Separable Problem Consider the binary classification
More informationMathematics Ext 2. HSC 2014 Solutions. Suite 403, 410 Elizabeth St, Surry Hills NSW 2010 keystoneeducation.com.
Mathematics Ext HSC 4 Solutions Suite 43, 4 Elizabeth St, Surry Hills NSW info@keystoneeducation.com.au keystoneeducation.com.au Mathematics Extension : HSC 4 Solutions Contents Multiple Choice... 3 Question...
More informationSpecialist Mathematics 2019 v1.2
181314 Mensuration circumference of a circle area of a parallelogram CC = ππππ area of a circle AA = ππrr AA = h area of a trapezium AA = 1 ( + )h area of a triangle AA = 1 h total surface area of a cone
More informationCSE 473: Artificial Intelligence
CSE 473: Artificial Intelligence Hidden Markov Models Dieter Fox --- University of Washington [Most slides were created by Dan Klein and Pieter Abbeel for CS188 Intro to AI at UC Berkeley. All CS188 materials
More informationME5286 Robotics Spring 2017 Quiz 2
Page 1 of 5 ME5286 Robotics Spring 2017 Quiz 2 Total Points: 30 You are responsible for following these instructions. Please take a minute and read them completely. 1. Put your name on this page, any other
More informationRotational Motion. Chapter 10 of Essential University Physics, Richard Wolfson, 3 rd Edition
Rotational Motion Chapter 10 of Essential University Physics, Richard Wolfson, 3 rd Edition 1 We ll look for a way to describe the combined (rotational) motion 2 Angle Measurements θθ ss rr rrrrrrrrrrrrrr
More information1 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 informationRobotics. Lecture 4: Probabilistic Robotics. See course website for up to date information.
Robotics Lecture 4: Probabilistic Robotics See course website http://www.doc.ic.ac.uk/~ajd/robotics/ for up to date information. Andrew Davison Department of Computing Imperial College London Review: Sensors
More informationState Estimation and Motion Tracking for Spatially Diverse VLC Networks
State Estimation and Motion Tracking for Spatially Diverse VLC Networks GLOBECOM Optical Wireless Communications Workshop December 3, 2012 Anaheim, CA Michael Rahaim mrahaim@bu.edu Gregary Prince gbprince@bu.edu
More informationSensors for mobile robots
ROBOTICS 01PEEQW Basilio Bona DAUIN Politecnico di Torino Mobile & Service Robotics Sensors for Robotics 2 Sensors for mobile robots Sensors are used to perceive, analyze and understand the environment
More informationLecture 3. STAT161/261 Introduction to Pattern Recognition and Machine Learning Spring 2018 Prof. Allie Fletcher
Lecture 3 STAT161/261 Introduction to Pattern Recognition and Machine Learning Spring 2018 Prof. Allie Fletcher Previous lectures What is machine learning? Objectives of machine learning Supervised and
More informationAutonomous 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 informationMobile Robot Localization
Mobile Robot Localization 1 The Problem of Robot Localization Given a map of the environment, how can a robot determine its pose (planar coordinates + orientation)? Two sources of uncertainty: - observations
More informationKalman 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 informationProbabilistic Fundamentals in Robotics
Probabilistic Fundamentals in Robotics Probabilistic Models of Mobile Robots Robot localization Basilio Bona DAUIN Politecnico di Torino June 2011 Course Outline Basic mathematical framework Probabilistic
More informationSLAM for Ship Hull Inspection using Exactly Sparse Extended Information Filters
SLAM for Ship Hull Inspection using Exactly Sparse Extended Information Filters Matthew Walter 1,2, Franz Hover 1, & John Leonard 1,2 Massachusetts Institute of Technology 1 Department of Mechanical Engineering
More informationA Step Towards the Cognitive Radar: Target Detection under Nonstationary Clutter
A Step Towards the Cognitive Radar: Target Detection under Nonstationary Clutter Murat Akcakaya Department of Electrical and Computer Engineering University of Pittsburgh Email: akcakaya@pitt.edu Satyabrata
More informationUAV Navigation: Airborne Inertial SLAM
Introduction UAV Navigation: Airborne Inertial SLAM Jonghyuk Kim Faculty of Engineering and Information Technology Australian National University, Australia Salah Sukkarieh ARC Centre of Excellence in
More informationLinear Dynamical Systems
Linear Dynamical Systems Sargur N. srihari@cedar.buffalo.edu Machine Learning Course: http://www.cedar.buffalo.edu/~srihari/cse574/index.html Two Models Described by Same Graph Latent variables Observations
More informationSECTION 5: POWER FLOW. ESE 470 Energy Distribution Systems
SECTION 5: POWER FLOW ESE 470 Energy Distribution Systems 2 Introduction Nodal Analysis 3 Consider the following circuit Three voltage sources VV sss, VV sss, VV sss Generic branch impedances Could be
More informationCS 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 informationRobotics. Mobile Robotics. Marc Toussaint U Stuttgart
Robotics Mobile Robotics State estimation, Bayes filter, odometry, particle filter, Kalman filter, SLAM, joint Bayes filter, EKF SLAM, particle SLAM, graph-based SLAM Marc Toussaint U Stuttgart DARPA Grand
More informationSensors Fusion for Mobile Robotics localization. M. De Cecco - Robotics Perception and Action
Sensors Fusion for Mobile Robotics localization 1 Until now we ve presented the main principles and features of incremental and absolute (environment referred localization systems) could you summarize
More informationthe 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 informationFrom 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 informationIntroduction 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 informationTSOKOS READING ACTIVITY Section 7-2: The Greenhouse Effect and Global Warming (8 points)
IB PHYSICS Name: Period: Date: DEVIL PHYSICS BADDEST CLASS ON CAMPUS TSOKOS READING ACTIVITY Section 7-2: The Greenhouse Effect and Global Warming (8 points) 1. IB Assessment Statements for Topic 8.5.
More informationQuantum Mechanics. An essential theory to understand properties of matter and light. Chemical Electronic Magnetic Thermal Optical Etc.
Quantum Mechanics An essential theory to understand properties of matter and light. Chemical Electronic Magnetic Thermal Optical Etc. Fall 2018 Prof. Sergio B. Mendes 1 CHAPTER 3 Experimental Basis of
More information3. Mathematical Modelling
3. Mathematical Modelling 3.1 Modelling principles 3.1.1 Model types 3.1.2 Model construction 3.1.3 Modelling from first principles 3.2 Models for technical systems 3.2.1 Electrical systems 3.2.2 Mechanical
More informationEKF and SLAM. McGill COMP 765 Sept 18 th, 2017
EKF and SLAM McGill COMP 765 Sept 18 th, 2017 Outline News and information Instructions for paper presentations Continue on Kalman filter: EKF and extension to mapping Example of a real mapping system:
More informationMobile Robot Localization
Mobile Robot Localization 1 The Problem of Robot Localization Given a map of the environment, how can a robot determine its pose (planar coordinates + orientation)? Two sources of uncertainty: - observations
More informationWorksheets for GCSE Mathematics. Quadratics. mr-mathematics.com Maths Resources for Teachers. Algebra
Worksheets for GCSE Mathematics Quadratics mr-mathematics.com Maths Resources for Teachers Algebra Quadratics Worksheets Contents Differentiated Independent Learning Worksheets Solving x + bx + c by factorisation
More informationAngular Momentum, Electromagnetic Waves
Angular Momentum, Electromagnetic Waves Lecture33: Electromagnetic Theory Professor D. K. Ghosh, Physics Department, I.I.T., Bombay As before, we keep in view the four Maxwell s equations for all our discussions.
More informationRobotics 2 Target Tracking. Kai Arras, Cyrill Stachniss, Maren Bennewitz, Wolfram Burgard
Robotics 2 Target Tracking Kai Arras, Cyrill Stachniss, Maren Bennewitz, Wolfram Burgard Slides by Kai Arras, Gian Diego Tipaldi, v.1.1, Jan 2012 Chapter Contents Target Tracking Overview Applications
More informationF.4 Solving Polynomial Equations and Applications of Factoring
section F4 243 F.4 Zero-Product Property Many application problems involve solving polynomial equations. In Chapter L, we studied methods for solving linear, or first-degree, equations. Solving higher
More informationCharge carrier density in metals and semiconductors
Charge carrier density in metals and semiconductors 1. Introduction The Hall Effect Particles must overlap for the permutation symmetry to be relevant. We saw examples of this in the exchange energy in
More informationCS249: ADVANCED DATA MINING
CS249: ADVANCED DATA MINING Vector Data: Clustering: Part II Instructor: Yizhou Sun yzsun@cs.ucla.edu May 3, 2017 Methods to Learn: Last Lecture Classification Clustering Vector Data Text Data Recommender
More informationPhysical Pendulum, Torsion Pendulum
[International Campus Lab] Physical Pendulum, Torsion Pendulum Objective Investigate the motions of physical pendulums and torsion pendulums. Theory ----------------------------- Reference --------------------------
More informationCHAPTER 2 Special Theory of Relativity
CHAPTER 2 Special Theory of Relativity Fall 2018 Prof. Sergio B. Mendes 1 Topics 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 2.11 2.12 2.13 2.14 Inertial Frames of Reference Conceptual and Experimental
More informationKalman 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 informationLecture No. 1 Introduction to Method of Weighted Residuals. Solve the differential equation L (u) = p(x) in V where L is a differential operator
Lecture No. 1 Introduction to Method of Weighted Residuals Solve the differential equation L (u) = p(x) in V where L is a differential operator with boundary conditions S(u) = g(x) on Γ where S is a differential
More informationLocal Probabilistic Models: Continuous Variable CPDs
Local Probabilistic Models: Continuous Variable CPDs Sargur srihari@cedar.buffalo.edu 1 Topics 1. Simple discretizing loses continuity 2. Continuous Variable CPDs 3. Linear Gaussian Model Example of car
More informationGeneral Strong Polarization
General Strong Polarization Madhu Sudan Harvard University Joint work with Jaroslaw Blasiok (Harvard), Venkatesan Gurswami (CMU), Preetum Nakkiran (Harvard) and Atri Rudra (Buffalo) December 4, 2017 IAS:
More informationSLAM Techniques and Algorithms. Jack Collier. Canada. Recherche et développement pour la défense Canada. Defence Research and Development Canada
SLAM Techniques and Algorithms Jack Collier Defence Research and Development Canada Recherche et développement pour la défense Canada Canada Goals What will we learn Gain an appreciation for what SLAM
More informationParticle Filters; Simultaneous Localization and Mapping (Intelligent Autonomous Robotics) Subramanian Ramamoorthy School of Informatics
Particle Filters; Simultaneous Localization and Mapping (Intelligent Autonomous Robotics) Subramanian Ramamoorthy School of Informatics Recap: State Estimation using Kalman Filter Project state and error
More informationRobot Dynamics II: Trajectories & Motion
Robot Dynamics II: Trajectories & Motion Are We There Yet? METR 4202: Advanced Control & Robotics Dr Surya Singh Lecture # 5 August 23, 2013 metr4202@itee.uq.edu.au http://itee.uq.edu.au/~metr4202/ 2013
More informationBig Bang Planck Era. This theory: cosmological model of the universe that is best supported by several aspects of scientific evidence and observation
Big Bang Planck Era Source: http://www.crystalinks.com/bigbang.html Source: http://www.odec.ca/index.htm This theory: cosmological model of the universe that is best supported by several aspects of scientific
More informationApproximate Second Order Algorithms. Seo Taek Kong, Nithin Tangellamudi, Zhikai Guo
Approximate Second Order Algorithms Seo Taek Kong, Nithin Tangellamudi, Zhikai Guo Why Second Order Algorithms? Invariant under affine transformations e.g. stretching a function preserves the convergence
More informationMath 171 Spring 2017 Final Exam. Problem Worth
Math 171 Spring 2017 Final Exam Problem 1 2 3 4 5 6 7 8 9 10 11 Worth 9 6 6 5 9 8 5 8 8 8 10 12 13 14 15 16 17 18 19 20 21 22 Total 8 5 5 6 6 8 6 6 6 6 6 150 Last Name: First Name: Student ID: Section:
More information2D Image Processing (Extended) Kalman and particle filter
2D Image Processing (Extended) Kalman and particle filter Prof. Didier Stricker Dr. Gabriele Bleser Kaiserlautern University http://ags.cs.uni-kl.de/ DFKI Deutsches Forschungszentrum für Künstliche Intelligenz
More informationRobot Localisation. Henrik I. Christensen. January 12, 2007
Robot Henrik I. Robotics and Intelligent Machines @ GT College of Computing Georgia Institute of Technology Atlanta, GA hic@cc.gatech.edu January 12, 2007 The Robot Structure Outline 1 2 3 4 Sum of 5 6
More information10.4 Controller synthesis using discrete-time model Example: comparison of various controllers
10. Digital 10.1 Basic principle of digital control 10.2 Digital PID controllers 10.2.1 A 2DOF continuous-time PID controller 10.2.2 Discretisation of PID controllers 10.2.3 Implementation and tuning 10.3
More informationProbabilistic Fundamentals in Robotics. DAUIN Politecnico di Torino July 2010
Probabilistic Fundamentals in Robotics Probabilistic Models of Mobile Robots Robot localization Basilio Bona DAUIN Politecnico di Torino July 2010 Course Outline Basic mathematical framework Probabilistic
More informationPhoton Interactions in Matter
Radiation Dosimetry Attix 7 Photon Interactions in Matter Ho Kyung Kim hokyung@pusan.ac.kr Pusan National University References F. H. Attix, Introduction to Radiological Physics and Radiation Dosimetry,
More informationPresenter: Siu Ho (4 th year, Doctor of Engineering) Other authors: Dr Andy Kerr, Dr Avril Thomson
The development and evaluation of a sensor-fusion and adaptive algorithm for detecting real-time upper-trunk kinematics, phases and timing of the sit-to-stand movements in stroke survivors Presenter: Siu
More information10.4 The Cross Product
Math 172 Chapter 10B notes Page 1 of 9 10.4 The Cross Product The cross product, or vector product, is defined in 3 dimensions only. Let aa = aa 1, aa 2, aa 3 bb = bb 1, bb 2, bb 3 then aa bb = aa 2 bb
More informationCharged-Particle Interactions in Matter
Radiation Dosimetry Attix 8 Charged-Particle Interactions in Matter Ho Kyung Kim hokyung@pusan.ac.kr Pusan National University References F. H. Attix, Introduction to Radiological Physics and Radiation
More informationRobots Autónomos. Depto. CCIA. 2. Bayesian Estimation and sensor models. Domingo Gallardo
Robots Autónomos 2. Bayesian Estimation and sensor models Domingo Gallardo Depto. CCIA http://www.rvg.ua.es/master/robots References Recursive State Estimation: Thrun, chapter 2 Sensor models and robot
More informationNational 5 Mathematics. Practice Paper E. Worked Solutions
National 5 Mathematics Practice Paper E Worked Solutions Paper One: Non-Calculator Copyright www.national5maths.co.uk 2015. All rights reserved. SQA Past Papers & Specimen Papers Working through SQA Past
More informationA new procedure for sensitivity testing with two stress factors
A new procedure for sensitivity testing with two stress factors C.F. Jeff Wu Georgia Institute of Technology Sensitivity testing : problem formulation. Review of the 3pod (3-phase optimal design) procedure
More informationUsing the Kalman Filter for SLAM AIMS 2015
Using the Kalman Filter for SLAM AIMS 2015 Contents Trivial Kinematics Rapid sweep over localisation and mapping (components of SLAM) Basic EKF Feature Based SLAM Feature types and representations Implementation
More informationTSRT14: Sensor Fusion Lecture 9
TSRT14: Sensor Fusion Lecture 9 Simultaneous localization and mapping (SLAM) Gustaf Hendeby gustaf.hendeby@liu.se TSRT14 Lecture 9 Gustaf Hendeby Spring 2018 1 / 28 Le 9: simultaneous localization and
More informationModule 7 (Lecture 25) RETAINING WALLS
Module 7 (Lecture 25) RETAINING WALLS Topics Check for Bearing Capacity Failure Example Factor of Safety Against Overturning Factor of Safety Against Sliding Factor of Safety Against Bearing Capacity Failure
More informationL06. LINEAR KALMAN FILTERS. NA568 Mobile Robotics: Methods & Algorithms
L06. LINEAR KALMAN FILTERS NA568 Mobile Robotics: Methods & Algorithms 2 PS2 is out! Landmark-based Localization: EKF, UKF, PF Today s Lecture Minimum Mean Square Error (MMSE) Linear Kalman Filter Gaussian
More informationSECTION 7: STEADY-STATE ERROR. ESE 499 Feedback Control Systems
SECTION 7: STEADY-STATE ERROR ESE 499 Feedback Control Systems 2 Introduction Steady-State Error Introduction 3 Consider a simple unity-feedback system The error is the difference between the reference
More informationPHY103A: Lecture # 4
Semester II, 2017-18 Department of Physics, IIT Kanpur PHY103A: Lecture # 4 (Text Book: Intro to Electrodynamics by Griffiths, 3 rd Ed.) Anand Kumar Jha 10-Jan-2018 Notes The Solutions to HW # 1 have been
More informationMobile Robots Localization
Mobile Robots Localization Institute for Software Technology 1 Today s Agenda Motivation for Localization Odometry Odometry Calibration Error Model 2 Robotics is Easy control behavior perception modelling
More informationLecture 6. Notes on Linear Algebra. Perceptron
Lecture 6. Notes on Linear Algebra. Perceptron COMP90051 Statistical Machine Learning Semester 2, 2017 Lecturer: Andrey Kan Copyright: University of Melbourne This lecture Notes on linear algebra Vectors
More informationWork, Energy, and Power. Chapter 6 of Essential University Physics, Richard Wolfson, 3 rd Edition
Work, Energy, and Power Chapter 6 of Essential University Physics, Richard Wolfson, 3 rd Edition 1 With the knowledge we got so far, we can handle the situation on the left but not the one on the right.
More informationIntegrated Attitude Determination and Control System via Magnetic Measurements and Actuation
Proceedings of the th WSEAS International Conference on Automatic Control, Modelling and Simulation Integrated Attitude Determination and Control System via Magnetic Measurements and Actuation MOHAMMAD
More informationESTIMATOR STABILITY ANALYSIS IN SLAM. Teresa Vidal-Calleja, Juan Andrade-Cetto, Alberto Sanfeliu
ESTIMATOR STABILITY ANALYSIS IN SLAM Teresa Vidal-Calleja, Juan Andrade-Cetto, Alberto Sanfeliu Institut de Robtica i Informtica Industrial, UPC-CSIC Llorens Artigas 4-6, Barcelona, 88 Spain {tvidal, cetto,
More informationThe Elasticity of Quantum Spacetime Fabric. Viorel Laurentiu Cartas
The Elasticity of Quantum Spacetime Fabric Viorel Laurentiu Cartas The notion of gravitational force is replaced by the curvature of the spacetime fabric. The mass tells to spacetime how to curve and the
More informationSimultaneous Localization and Map Building Using Natural features in Outdoor Environments
Simultaneous Localization and Map Building Using Natural features in Outdoor Environments Jose Guivant, Eduardo Nebot, Hugh Durrant Whyte Australian Centre for Field Robotics Department of Mechanical and
More informationSimultaneous Localization and Mapping (SLAM) Corso di Robotica Prof. Davide Brugali Università degli Studi di Bergamo
Simultaneous Localization and Mapping (SLAM) Corso di Robotica Prof. Davide Brugali Università degli Studi di Bergamo Introduction SLAM asks the following question: Is it possible for an autonomous vehicle
More informationPartially Observable Markov Decision Processes (POMDPs)
Partially Observable Markov Decision Processes (POMDPs) Sachin Patil Guest Lecture: CS287 Advanced Robotics Slides adapted from Pieter Abbeel, Alex Lee Outline Introduction to POMDPs Locally Optimal Solutions
More informationCIS 390 Fall 2016 Robotics: Planning and Perception Final Review Questions
CIS 390 Fall 2016 Robotics: Planning and Perception Final Review Questions December 14, 2016 Questions Throughout the following questions we will assume that x t is the state vector at time t, z t is the
More informationMarkov Models. CS 188: Artificial Intelligence Fall Example. Mini-Forward Algorithm. Stationary Distributions.
CS 88: Artificial Intelligence Fall 27 Lecture 2: HMMs /6/27 Markov Models A Markov model is a chain-structured BN Each node is identically distributed (stationarity) Value of X at a given time is called
More informationCHAPTER 5 Wave Properties of Matter and Quantum Mechanics I
CHAPTER 5 Wave Properties of Matter and Quantum Mechanics I 1 5.1 X-Ray Scattering 5.2 De Broglie Waves 5.3 Electron Scattering 5.4 Wave Motion 5.5 Waves or Particles 5.6 Uncertainty Principle Topics 5.7
More informationWorksheets for GCSE Mathematics. Algebraic Expressions. Mr Black 's Maths Resources for Teachers GCSE 1-9. Algebra
Worksheets for GCSE Mathematics Algebraic Expressions Mr Black 's Maths Resources for Teachers GCSE 1-9 Algebra Algebraic Expressions Worksheets Contents Differentiated Independent Learning Worksheets
More information10.1 Three Dimensional Space
Math 172 Chapter 10A notes Page 1 of 12 10.1 Three Dimensional Space 2D space 0 xx.. xx-, 0 yy yy-, PP(xx, yy) [Fig. 1] Point PP represented by (xx, yy), an ordered pair of real nos. Set of all ordered
More informationADAPTIVE NEURAL NETWORK CONTROLLER DESIGN FOR BLENDED-WING UAV WITH COMPLEX DAMAGE
ADAPTIVE NEURAL NETWORK CONTROLLER DESIGN FOR BLENDED-WING UAV WITH COMPLEX DAMAGE Kijoon Kim*, Jongmin Ahn**, Seungkeun Kim*, Jinyoung Suk* *Chungnam National University, **Agency for Defense and Development
More information1 Introduction. Systems 2: Simulating Errors. Mobile Robot Systems. System Under. Environment
Systems 2: Simulating Errors Introduction Simulating errors is a great way to test you calibration algorithms, your real-time identification algorithms, and your estimation algorithms. Conceptually, the
More informationDistributed Intelligent Systems W4 An Introduction to Localization Methods for Mobile Robots
Distributed Intelligent Systems W4 An Introduction to Localization Methods for Mobile Robots 1 Outline Positioning systems Indoor Outdoor Robot localization using proprioceptive sensors without uncertainties
More informationLesson 24: Using the Quadratic Formula,
, b ± b 4ac x = a Opening Exercise 1. Examine the two equation below and discuss what is the most efficient way to solve each one. A. 4xx + 5xx + 3 = xx 3xx B. cc 14 = 5cc. Solve each equation with the
More informationManipulators. Robotics. Outline. Non-holonomic robots. Sensors. Mobile Robots
Manipulators P obotics Configuration of robot specified by 6 numbers 6 degrees of freedom (DOF) 6 is the minimum number required to position end-effector arbitrarily. For dynamical systems, add velocity
More information2D Image Processing. Bayes filter implementation: Kalman filter
2D Image Processing Bayes filter implementation: Kalman filter Prof. Didier Stricker Dr. Gabriele Bleser Kaiserlautern University http://ags.cs.uni-kl.de/ DFKI Deutsches Forschungszentrum für Künstliche
More informationLecture 11. Kernel Methods
Lecture 11. Kernel Methods COMP90051 Statistical Machine Learning Semester 2, 2017 Lecturer: Andrey Kan Copyright: University of Melbourne This lecture The kernel trick Efficient computation of a dot product
More informationProf. Dr.-Ing. Armin Dekorsy Department of Communications Engineering. Stochastic Processes and Linear Algebra Recap Slides
Prof. Dr.-Ing. Armin Dekorsy Department of Communications Engineering Stochastic Processes and Linear Algebra Recap Slides Stochastic processes and variables XX tt 0 = XX xx nn (tt) xx 2 (tt) XX tt XX
More informationBIOE 198MI Biomedical Data Analysis. Spring Semester Lab 4: Introduction to Probability and Random Data
BIOE 98MI Biomedical Data Analysis. Spring Semester 209. Lab 4: Introduction to Probability and Random Data A. Random Variables Randomness is a component of all measurement data, either from acquisition
More informationCHAPTER 4 Structure of the Atom
CHAPTER 4 Structure of the Atom Fall 2018 Prof. Sergio B. Mendes 1 Topics 4.1 The Atomic Models of Thomson and Rutherford 4.2 Rutherford Scattering 4.3 The Classic Atomic Model 4.4 The Bohr Model of the
More informationRadial Basis Function (RBF) Networks
CSE 5526: Introduction to Neural Networks Radial Basis Function (RBF) Networks 1 Function approximation We have been using MLPs as pattern classifiers But in general, they are function approximators Depending
More informationModule 4 (Lecture 16) SHALLOW FOUNDATIONS: ALLOWABLE BEARING CAPACITY AND SETTLEMENT
Topics Module 4 (Lecture 16) SHALLOW FOUNDATIONS: ALLOWABLE BEARING CAPACITY AND SETTLEMENT 1.1 STRIP FOUNDATION ON GRANULAR SOIL REINFORCED BY METALLIC STRIPS Mode of Failure Location of Failure Surface
More information