FastSLAM: A Scalable Method for the Simultaneous Localization and Mapping Problem in Robotics / Edition 1

FastSLAM: A Scalable Method for the Simultaneous Localization and Mapping Problem in Robotics / Edition 1

by Michael Montemerlo, Sebastian Thrun
     
 

This monograph describes a new family of algorithms for the simultaneous localization and mapping problem in robotics (SLAM). SLAM addresses the problem of acquiring an environment map with a roving robot, while simultaneously localizing the robot relative to this map. This problem has received enormous attention in the robotics community in the past few years,

See more details below

Overview

This monograph describes a new family of algorithms for the simultaneous localization and mapping problem in robotics (SLAM). SLAM addresses the problem of acquiring an environment map with a roving robot, while simultaneously localizing the robot relative to this map. This problem has received enormous attention in the robotics community in the past few years, reaching a peak of popularity on the occasion of the DARPA Grand Challenge in October 2005, which was won by the team headed by the authors. The FastSLAM family of algorithms applies particle filters to the SLAM Problem, which provides new insights into the data association problem that is paramount in SLAM. The FastSLAM-type algorithms have enabled robots to acquire maps of unprecedented size and accuracy, in a number of robot application domains and have been successfully applied in different dynamic environments, including the solution to the problem of people tracking.

Product Details

ISBN-13:
9783540463993
Publisher:
Springer Berlin Heidelberg
Publication date:
03/05/2007
Series:
Springer Tracts in Advanced Robotics Series, #27
Edition description:
2007
Pages:
120
Product dimensions:
6.10(w) x 9.20(h) x 0.70(d)

Table of Contents

1 Introduction 1

1.1 Applications of SLAM 1

1.2 Joint Estimation 2

1.3 Posterior Estimation 3

1.4 The Extended Kalman Filter 5

1.4.1 Quadratic Complexity 5

1.4.2 Single-Hypothesis Data Association 6

1.5 Structure and Sparsity in SLAM 7

1.6 FastSLAM 8

1.6.1 Logarithmic Complexity 10

1.6.2 Multi-hypothesis Data Association 10

1.7 Outline 11

2 The SLAM Problem 13

2.1 Problem Definition 13

2.2 SLAM Posterior 15

2.3 SLAM as a Markov Chain 16

2.3.1 Bayes Filter Derivation 17

2.4 Extended Kalman Filtering 18

2.5 Scaling SLAM Algorithms 20

2.5.1 Submap Methods 20

2.5.2 Sparse Extended Information Filters 21

2.5.3 Thin Junction Trees 22

2.5.4 Covariance Intersection 22

2.5.5 Graphical Optimization Methods 22

2.6 Robust Data Association 23

2.6.1 Local Map Sequencing 24

2.6.2 Joint Compatibility Branch and Bound 24

2.6.3 Combined Constraint Data Association 25

2.6.4 Iterative Closest Point 25

2.6.5 Multiple Hypothesis Tracking 25

2.7 Comparison of FastSLAM to Existing Techniques 26

3 FastSLAM 1.0 27

3.1 Particle Filtering 27

3.2 Factored Posterior Representation 29

3.2.1 Proof of the FastSLAM Factorization 30

3.3 The FastSLAM 1.0 Algorithm 32

3.3.1 Sampling a New Pose 33

3.3.2 Updating the Landmark Estimates 35

3.3.3 Calculating Importance Weights 37

3.3.4 Importance Resampling 38

3.3.5 Robot Path Posterior Revisited 39

3.4 FastSLAM with Unknown Data Association 39

3.4.1 Data Association Uncertainty 39

3.4.2 Per-Particle Data Association 41

3.4.3 Adding New Landmarks 43

3.5 Summary of the FastSLAM Algorithm 44

3.6 FastSLAM Extensions 46

3.6.1 Greedy Mutual Exclusion 46

3.6.2 Feature Elimination UsingNegative Evidence 47

3.7 Log(N) FastSLAM 48

3.7.1 Garbage Collection 50

3.7.2 Unknown Data Association 51

3.8 Experimental Results 51

3.8.1 Victoria Park 52

3.8.2 Comparison of FastSLAM and the EKF 56

3.8.3 Ambiguous Data Association 59

3.8.4 Sample Impoverishment 59

3.9 Summary 62

4 FastSLAM 2.0 63

4.1 Sample Impoverishment 63

4.2 FastSLAM 2.0 65

4.2.1 The New Proposal Distribution 66

4.2.2 Calculating the Importance Weights 69

4.2.3 FastSLAM 2.0 Overview 71

4.2.4 Handling Simultaneous Observations 71

4.3 FastSLAM 2.0 Convergence 74

4.3.1 Convergence Proof 75

4.4 Experimental Results 79

4.4.1 FastSLAM 1.0 Versus FastSLAM 2.0 79

4.4.2 One Particle FastSLAM 2.0 81

4.4.3 Scaling Performance 83

4.4.4 Loop Closing 83

4.4.5 Convergence Speed 85

4.5 Grid-Based FastSLAM 87

4.6 Summary 90

5 Dynamic Environments 91

5.1 SLAM with Dynamic Landmarks 92

5.1.1 Derivation of the Bayes Filter with Dynamic Objects 93

5.1.2 Factoring the Dynamic SLAM Problem 95

5.2 Simultaneous Localization and People Tracking 96

5.2.1 Comparison with Prior Work 97

5.3 FastSLAP Implementation 97

5.3.1 Scan-Based Data Association 98

5.3.2 Measurement Model 100

5.3.3 Motion Model 101

5.3.4 Model Selection 101

5.4 Experimental Results 102

5.4.1 Tracking and Model Selection Accuracy 102

5.4.2 Global Uncertainty 103

5.4.3 Intelligent Following Behavior 103

5.5 Summary 105

6 Conclusions 107

6.1 Conclusion 107

6.1.1 Limitations of FastSLAM 108

6.2 Future Work 109

References 111

Index 117

Read More

Customer Reviews

Average Review:

Write a Review

and post it to your social network

     

Most Helpful Customer Reviews

See all customer reviews >