Multimodal Perception and Secure State Estimation for Robotic Mobility Platforms - Liu, Xinghua

. .

Résumé :
..

Biographie:
..

Sommaire:

About the Authors xii

Preface xiv

1 Introduction 1

1.1 Background and Motivation 1

1.2 Multimodal Pose Estimation for Vehicle Navigation 2

1.2.1 Multi-Senor Pose Estimation 2

1.2.2 Pose Estimation with Constraints 4

1.2.3 Research Focus in Multimodal Pose Estimation 5

1.3 Secure Estimation 7

1.3.1 Secure State Estimation under Cyber Attacks 7

1.3.2 Secure Pose Estimation for Autonomous Vehicles 8

1.4 Contributions and Organization 9

Part I Multimodal Perception in Vehicle Pose Estimation 13

2 Heading Reference-Assisted Pose Estimation 15

2.1 Preliminaries 16

2.1.1 Stereo Visual Odometry 16

2.1.2 Heading Reference Sensors 17

2.1.3 Graph Optimization on a Manifold 17

2.2 Abstraction Model of Measurement with a Heading Reference 19

2.2.1 Loosely Coupled Model 19

2.2.2 Tightly Coupled Model 20

2.2.3 Structure of the Abstraction Model 22

2.2.4 Vertex Removal in the Abstraction Model 22

2.3 Heading Reference-Assisted Pose Estimation (HRPE) 24

2.3.1 Initialization 24

2.3.2 Graph Optimization 24

2.3.3 Maintenance of the Dynamic Graph 26

2.4 Simulation Studies 26

2.4.1 Accuracy with Respect to Heading Measurement Error 28

2.4.2 Accuracy with Respect to Sliding Window Size 28

2.4.3 Time Consumption with Respect to Sliding Window Size 28

2.5 Experimental Results 31

2.5.1 Experimental Platform 31

2.5.2 Pose Estimation Performance 33

2.5.3 Real-Time Performance 34

2.6 Conclusion 36

3 Road-Constrained Localization Using Cloud Models 37

3.1 Preliminaries 38

3.1.1 Scaled Measurement Equations for Visual Odometry 38

3.1.2 Cloud Models 39

3.1.3 Uniform Gaussian Distribution (UGD) 39

3.1.4 Gaussian-Gaussian Distribution (GGD) 42

3.2 Map-Assisted Ground Vehicle Localization 43

3.2.1 Measurement Representation with UGD 44

3.2.2 Shape Matching Between Map and Particles 45

3.2.3 Particle Resampling and Parameter Estimation 46

3.2.4 Framework Extension to Other Cloud Models 47

3.3 Experimental Validation on UGD 47

3.3.1 Configurations 47

3.3.2 Localization with Stereo Visual Odometry 48

3.3.3 Localization with Monocular Visual Odometry 49

3.3.4 Scale Estimation Results 52

3.3.5 Weighting Function Balancing 52

3.4 Experimental Validation on GGD 54

3.4.1 Experiments on KITTI 55

3.4.2 Experiments on the Self-Collected Dataset 61

3.5 Conclusion 63

4 GPS/Odometry/Map Fusion for Vehicle Positioning Using Potential Functions 65

4.1 Potential Wells and Potential Trenches 66

4.1.1 Potential Function Creation 67

4.1.2 Minimum Searching 71

4.2 Potential-Function-Based Fusion for Vehicle Positioning 74

4.2.1 Information Sources and Sensors 74

4.2.2 Potential Representation 76

4.2.3 Road-Switching Strategy 76

4.3 Experimental Results 78

4.3.1 Quantitative Results 78

4.3.2 Qualitative Evaluation 80

4.4 Conclusion 84

5 Multi-Sensor Geometric Pose Estimation 85

5.1 Preliminaries 86

5.1.1 Distance on Riemannian Manifolds 86

5.1.2 Probabilistic Distribution on Riemannian Manifolds 87

5.2 Geometric Pose Estimation Using Dynamic Potential Fields 88

5.2.1 State Space and Measurement Space 88

5.2.2 Dynamic Potential Fields on Manifolds 90

5.2.3 DPF-Based Information Fusion 91

5.2.4 Approximation of Geometric Pose Estimation 95

5.3 VO-Heading-Map Pose Esti...