Robot navigation from nature : simultaneous localisation, mapping, and path planning based on hippocampal models

At the dawn of the new millennium, robotics is undergoing a major transformation in scope and dimension. From a largely dominant industrial focus, robotics is rapidly expanding into the challenges of unstructured environments. Interacting with, assi- ing, serving, and exploring with humans, the emerging robots will increasingly touch people and their lives. The goal of the new series of Springer Tracts in Advanced Robotics (STAR) is to bring, in a timely fashion, the latest advances and developments in robotics on the basis of their significance and quality. It is our hope that the wider dissemination of research developments will stimulate more exchanges and collaborations among the research community and contribute to further advancement of this rapidly growing field. The monograph written by Michael Milford is yet another volume in the series devoted to one of the hottest research topics in the latest few years, namely Simul- neous Localization and Map Building (SLAM). The contents expand the author's doctoral dissertation and describe the development of a robot mapping and navigation system inspired by models of the neural mechanisms underlying spatial navigation in the rodent hippocampus. One unique merit of the book lies in its truly interdisciplinary flavour, addressing a link between biology and artificial robotic systems that has been open for many years. To the best of my knowledge, this is the most thorough attempt to marry biological navigation in rodents and similar, with robotic navigation and SLAM. A very fine addition to our STAR series!

List of Figures	p. xv
List of Abbreviations	p. xix
1 Introduction	p. 1
1.1 Mobile Robots	p. 1
1.2 Simultaneous Localisation and Mapping	p. 3
1.3 Exploration, Goal Navigation and Adapting to Change	p. 5
1.4 Practical Performance from a Biological Model	p. 6
1.5 Book Outline	p. 6
2 Mapping and Navigation	p. 9
2.1 The Mapping and Navigation Problem	p. 10
2.1.1 Localisation and Mapping	p. 10
2.1.2 Slam: The Chicken and the Egg Problem	p. 11
2.1.3 Dealing with Uncertainty	p. 11
2.1.4 Exploring Unknown Environments	p. 12
2.1.5 Navigating to Goals	p. 12
2.1.6 Learning and Coping with Change	p. 13
3 Robotic Mapping Methods	p. 15
3.1 Probabilistic Mapping Algorithms	p. 15
3.1.1 Kalman Filter Methods	p. 15
3.1.2 Expectation Maximisation Methods	p. 17
3.1.3 Particle Filter Methods	p. 18
3.2 Topological Mapping Methods	p. 21
3.3 Exploration, Navigation, and Dealing with Change	p. 22
3.3.1 Exploration	p. 23
3.3.2 Navigating to Goals	p. 25
3.3.3 Dealing with Dynamic Environments	p. 26
3.4 Discussion	p. 28
4 Biological Navigation Systems	p. 29
4.1 Rodents and the Cognitive Map	p. 29
4.1.1 Head Direction and Place Cells	p. 31
4.1.2 Exploration, Navigation, and Dealing with Change	p. 33
4.2 Other Animals and Insects	p. 34
4.2.1 Bees	p. 34
4.2.2 Ants	p. 36
4.2.3 Primates	p. 36
4.2.4 Humans	p. 38
4.3 Discussion	p. 39
5 Emulating Nature: Models of Hippocampus	p. 41
5.1 Head Direction and Place Cells - State of the Art	p. 41
5.1.1 Attractor Networks	p. 41
5.1.2 Path Integration	p. 42
5.1.3 Head Direction Correction Using Allothetic Information	p. 44
5.1.4 Place Cells - State of the Art	p. 45
5.1.5 Place Cells Through Allothetic Cues	p. 45
5.1.6 Place Cells Through Ideothetic Information	p. 47
5.1.7 Navigation	p. 51
5.2 Discussion	p. 53
6 Robotic or Bio-inspired: A Comparison	p. 55
6.1 Robustness Versus Accuracy	p. 55
6.2 Map Friendliness Versus Map Usability	p. 56
6.3 Sensory Differences	p. 57
6.4 Capability in Real World Environments	p. 58
6.5 One Solution	p. 59
7 Pilot Study of a Hippocampal Model	p. 61
7.1 Robot and Environment	p. 61
7.2 Complete Model Structure	p. 63
7.3 A Model of Spatial Orientation	p. 64
7.3.1 Representing Orientation	p. 64
7.3.2 Learning Allothetic Cues	p. 66
7.3.3 Re-localisation Using Allothetic Cues	p. 67
7.3.4 Internal Dynamics	p. 68
7.3.5 Path Integration Using Ideothetic Information	p. 69
7.4 Model Performance	p. 70
7.4.1 Experiment 1: Path Integration Calibration	p. 70
7.4.2 Experiment 2: Localisation and Mapping in 1D	p. 71
7.5 A Model of Spatial Location	p. 74
7.5.1 Representing Location	p. 74
7.5.2 Learning Allothetic Cues	p. 75
7.5.3 Re-Localisation Using Allothetic Cues	p. 75
7.5.4 Internal Dynamics	p. 76
7.5.5 Path Integration Using Ideothetic Information	p. 78
7.6 Model Performance	p. 79
7.6.1 Experiment 3: Localisation and Mapping in 2D	p. 79
7.7 Discussion and Summary	p. 81
7.7.1 Comparison to Biological Systems	p. 83
7.7.2 Comparison to Other Models	p. 84
7.7.3 Conclusion	p. 86
8 RatSLAM: An Extended Hippocampal Model	p. 87
8.1 A Model of Spatial Pose	p. 87
8.1.1 Complete Model Structure	p. 87
8.1.2 Biological Evidence for Pose Cells	p. 87
8.1.3 Representing Pose	p. 89
8.1.4 Internal Dynamics	p. 90
8.1.5 Learning Visual Scenes	p. 91
8.1.6 Re-localising Using Familiar Visual Scenes	p. 92
8.1.7 Intuitive Path Integration	p. 93
8.2 Generation of Local View	p. 94
8.2.1 Sum of Absolute Differences Module	p. 94
8.2.2 Image Histograms	p. 96
8.3 Visualising SLAM in a Hippocampal Model	p. 99
8.4 Slam in Indoor and Outdoor Environments	p. 101
8.4.1 Experiment 4: Slam with Artificial Landmarks	p. 101
8.4.2 Experiment 5: Slam in a Loop Environment	p. 104
8.4.3 Experiment 6: Slam in an Office Building	p. 107
8.4.4 Experiment 7: Slam in Outdoor Environments	p. 112
8.4.5 Path Integration only Performance	p. 113
8.4.6 Slam Results	p. 113
8.5 Summary and Discussion	p. 115
8.5.1 RatSlam Requirements	p. 115
8.5.2 The Nature of RatSlam Representations	p. 115
9 Goal Memory: A Pilot Study	p. 117
9.1 Enabling Goal Recall Using RatSlam	p. 117
9.2 Learning	p. 118
9.3 Recall	p. 119
9.3.1 Experiment 8: Small Environment Goal Recall	p. 120
9.3.2 Goal Recall Results	p. 121
9.3.3 Experiment 9: Large Environment Goal Recall	p. 124
9.3.4 Goal Recall Results	p. 125
9.4 Summary and Discussion	p. 126
9.4.1 Creating Maps Suited to Goal Recall	p. 127
10 Extending RatSlam: The Experience Mapping Algorithm	p. 129
10.1 A Map Made of Experiences	p. 129
10.2 Linking Experiences: Spatially, Temporally, Behaviourally	p. 131
10.3 Map Correction	p. 133
10.4 Map Adaptation and Long Term Maintenance	p. 134
10.5 Indoor Experience Mapping Results	p. 136
10.5.1 Experiment 10: Large Pose Cell Representation	p. 137
10.5.2 Experiment 11: Small Pose Cell Representation	p. 139
10.6 Experiment 12: Outdoor Experience Mapping	p. 140
10.7 Summary and Discussion	p. 142
11 Exploration, Goal Recall, and Adapting to Change	p. 145
11.1 Exploring Efficiently	p. 145
11.1.1 Experimental Evaluation	p. 147
11.1.2 Discussion	p. 147
11.2 Recalling Routes Using a Temporal Map	p. 149
11.2.1 Temporal Map Creation	p. 149
11.2.2 Route Planning	p. 150
11.2.3 Behaviour Arbitration	p. 151
11.2.4 Route Loss Recovery	p. 153
11.3 Slam and Navigation in a Static Environment	p. 153
11.3.1 Experiment 13: Goal Recall with Minor Pose Collisions	p. 154
11.3.2 Experiment 14: Goal Recall with Major Pose Collisions	p. 156
11.3.3 Discussion	p. 156
11.4 Adapting to Environment Change	p. 158
11.4.1 Experiment 15: Indoor Map Adaptation	p. 158
11.4.2 Results	p. 159
11.5 Discussion	p. 160
11.5.1 Conclusion	p. 161
12 Discussion	p. 163
12.1 Book Summary	p. 163
12.1.1 Mapping and Navigation	p. 163
12.1.2 Pilot Study of a Hippocampal Model	p. 164
12.1.3 RatSlam: An Extended Hippocampal Model	p. 164
12.1.4 Goal Memory: A Pilot Study	p. 165
12.1.5 Extending RatSlam: Experience Mapping	p. 165
12.1.6 Exploring, Goal Recall, and Adapting to Change	p. 165
12.2 Contributions	p. 166
12.2.1 Comparative Review of Robotic and Biological Systems	p. 166
12.2.2 Performance Evaluation of Hippocampal Models	p. 167
12.2.3 Implementation of an Extended Hippocampal Model	p. 167
12.2.4 An Experience Mapping Algorithm	p. 167
12.2.5 An Integrated Approach to Mapping and Navigation	p. 168
12.3 Future Mapping and Navigation Research	p. 168
12.4 Grid Cells	p. 170
12.5 Conclusion	p. 170
Appendix A The Movement Behaviours	p. 173
Search of Local Space	p. 173
Pathway Identification	p. 173
Velocity Commands	p. 176
References	p. 179
List of Reproduced Figures	p. 187
Index	p. 191

Available:*

On Order

Summary

Summary

Table of Contents