Learning for Adaptive & Reactive Robot Control

by Billard, Mirrazavi

| ISBN: 9780262367028 | Copyright 2022

Click here to preview

Instructor Requests

Digital Exam/Desk Copy Print Desk Copy Ancillaries
Tabs
Expand/Collapse All
Cover (pg. i)
I Introduction (pg. 1)
1 Using and Learning Dynamical Systemsfor Robot Control—Overview (pg. 3)
1.1 Prerequisites and Additional Material (pg. 3)
1.2 Trajectory Planning under Uncertainty (pg. 4)
1.3 Computing Paths with DSs (pg. 9)
1.4 Learning a Control Law to Plan Paths Automatically (pg. 13)
1.5 Learning How to Combine Control Laws (pg. 14)
1.6 Modifying a Control Law through Learning (pg. 15)
1.7 Coupling DSs (pg. 18)
1.8 Generating and Learning Compliant Control with DSs (pg. 20)
1.9 Control Architectures (pg. 22)
2 Gathering Data for Learning (pg. 27)
2.1 Approaches to Generate Data (pg. 27)
2.2 Interfaces for Teaching Robots (pg. 29)
2.3 Desiderata for the Data (pg. 34)
2.4 Teaching a Robot How to Play Golf (pg. 36)
2.5 Gathering Data from Optimal Control (pg. 40)
3 Learning a Control Law (pg. 45)
3.1 Preliminaries (pg. 46)
3.2 Nonlinear DSs as a Mixture of Linear Systems (pg. 55)
3.3 Learning Stable, Nonlinear DSs (pg. 57)
3.4 Learning Stable, Highly Nonlinear DSs (pg. 76)
3.5 Learning Stable, Second-Order DSs (pg. 103)
3.6 Conclusion (pg. 109)
II Learning a Controller (pg. 43)
4 Learning Multiple Control Laws (pg. 111)
4.1 Combining Control Laws through State-Space Partitioning (pg. 111)
4.2 Learning of DSs with Bifurcations (pg. 121)
5 Learning Sequences of Control Laws (pg. 131)
5.1 Learning Locally Active Globally Stable Dynamical Systems (pg. 133)
5.2 Learning Sequences of LPV-DS with Hidden Markov Models (pg. 154)
III Coupling and Modulating Controllers (pg. 173)
6 Coupling and Synchronizing Controllers (pg. 175)
6.1 Preliminaries (pg. 176)
6.2 Coupling Two Linear DSs (pg. 177)
6.3 Coupling Arm-Hand Movement1 (pg. 180)
6.4 Coupling Eye-Hand-Arm Movements4 (pg. 189)
7 Reaching for and Adapting to Moving Objects (pg. 195)
7.1 How to Reach for a Moving Object (pg. 196)
7.2 Unimanual Reaching for a Fixed Small Object (pg. 198)
7.3 Unimanual Reaching for a Moving Small Object (pg. 202)
7.4 Robotic Implementation (pg. 205)
7.5 Bimanual Reaching for a Moving Large Object (pg. 209)
7.6 Robotic Implementation (pg. 213)
8 Adapting and Modulating an Existing Control Law (pg. 219)
8.1 Preliminaries (pg. 219)
8.2 Learning an Internal Modulation (pg. 223)
8.3 Learning an External Modulation4 (pg. 230)
8.4 Modulation to Transit from Free Space to Contact (pg. 236)
9 Obstacle Avoidance (pg. 245)
9.1 Obstacle Avoidance: Formalism (pg. 246)
9.2 Self-Collision, Joint-Level Obstacle Avoidance (pg. 257)
IV Compliant and Force Control with Dynamical Systems (pg. 267)
10 Compliant Control (pg. 269)
10.1 When and Why Should a Robot Be Compliant? (pg. 269)
10.2 Compliant Motion Generators (pg. 273)
10.3 Learning the Desired Impedance Profiles (pg. 285)
10.4 Passive Interaction Control with DSs (pg. 287)
11 Force Control (pg. 295)
11.1 Motion and Force Generation in Contact Tasks with DSs (pg. 295)
12 Conclusion and Outlook (pg. 303)
AppA-B (pg. 305)
V Appendices (pg. 305)
A Background on Dynamical Systems Theory (pg. 307)
A.1 Dynamical Systems (pg. 307)
A.2 Visualization of Dynamical Systems (pg. 308)
A.3 Linear and Nonlinear Dynamical Systems (pg. 308)
A.4 Stability Definitions (pg. 309)
A.5 Stability Analysis and Lyapunov Stability (pg. 311)
A.6 Energy Conservation and Passivity (pg. 312)
A.7 Limit Cycles (pg. 313)
A.8 Bifurcations (pg. 314)
B Background on Machine Learning (pg. 315)
B.1 Machine Learning Problems (pg. 315)
B.2 Metrics (pg. 316)
B.3 Gaussian Mixture Models (pg. 319)
B.4 Support Vector Machines (pg. 337)
B.5 Gaussian Processes Regression (pg. 348)
AppC-D (pg. 357)
C Background on Robot Control (pg. 357)
C.1 Multi-rigid Body Dynamics (pg. 357)
C.2 Motion Control (pg. 357)
D Proofs and Derivations (pg. 361)
D.1 Proofs and Derivations for Chapter 3 (pg. 361)
D.2 Proofs and Derivations for Chapter 4 (pg. 362)
D.3 Proofs and Derivations for Chapter 5 (pg. 363)
D.4 Proofs and Derivations for Chapter 9 (pg. 373)
Endnotes (pg. 379)
Bibliography (pg. 383)
Index (pg. 391)
Contents (pg. vii)
Preface (pg. xiii)
Notation (pg. xix)
I. Introduction (pg. 1)
1. Using and Learning Dynamical Systems for Robot Control—Overview (pg. 3)
1.1 Prerequisites and Additional Material (pg. 3)
1.2 Trajectory Planning under Uncertainty (pg. 4)
1.3 Computing Paths with DSs (pg. 9)
1.4 Learning a Control Law to Plan Paths Automatically (pg. 13)
1.5 Learning How to Combine Control Laws (pg. 14)
1.6 Modifying a Control Law through Learning (pg. 15)
1.7 Coupling DSs (pg. 18)
1.8 Generating and Learning Compliant Control with DSs (pg. 20)
1.9 Control Architectures (pg. 22)
2. Gathering Data for Learning (pg. 27)
2.1 Approaches to Generate Data (pg. 27)
2.2 Interfaces for Teaching Robots (pg. 29)
2.3 Desiderata for the Data (pg. 34)
2.4 Teaching a Robot How to Play Golf (pg. 36)
2.5 Gathering Data from Optimal Control (pg. 40)
II. Learning a Controller (pg. 43)
3. Learning a Control Law (pg. 45)
3.1 Preliminaries (pg. 46)
3.2 Nonlinear DSs as a Mixture of Linear Systems (pg. 55)
3.3 Learning Stable, Nonlinear DSs (pg. 57)
3.4 Learning Stable, Highly Nonlinear DSs (pg. 76)
3.5 Learning Stable, Second-Order DSs (pg. 103)
3.6 Conclusion (pg. 109)
4. Learning Multiple Control Laws (pg. 111)
4.1 Combining Control Laws through State-Space Partitioning (pg. 111)
4.2 Learning of DSs with Bifurcations (pg. 121)
5. Learning Sequences of Control Laws (pg. 131)
5.1 Learning Locally Active Globally Stable Dynamical Systems (pg. 133)
5.2 Learning Sequences of LPV-DS with Hidden Markov Models (pg. 154)
III. Coupling and Modulating Controllers (pg. 173)
6. Coupling and Synchronizing Controllers (pg. 175)
6.1 Preliminaries (pg. 176)
6.2 Coupling Two Linear DSs (pg. 177)
6.3 Coupling Arm-Hand Movement (pg. 180)
6.4 Coupling Eye-Hand-Arm Movements (pg. 189)
7. Reaching for and Adapting to Moving Objects (pg. 195)
7.1 How to Reach for a Moving Object (pg. 196)
7.2 Unimanual Reaching for a Fixed Small Object (pg. 198)
7.3 Unimanual Reaching for a Moving Small Object (pg. 202)
7.4 Robotic Implementation (pg. 205)
7.5 Bimanual Reaching for a Moving Large Object (pg. 209)
7.6 Robotic Implementation (pg. 213)
8. Adapting and Modulating an Existing Control Law (pg. 219)
8.1 Preliminaries (pg. 219)
8.2 Learning an Internal Modulation (pg. 223)
8.3 Learning an External Modulation (pg. 230)
8.4 Modulation to Transit from Free Space to Contact (pg. 236)
9. Obstacle Avoidance (pg. 245)
9.1 Obstacle Avoidance: Formalism (pg. 246)
9.2 Self-Collision, Joint-Level Obstacle Avoidance (pg. 257)
IV. Compliant and Force Control with Dynamical Systems (pg. 267)
10. Compliant Control (pg. 269)
10.1 When and Why Should a Robot Be Compliant? (pg. 269)
10.2 Compliant Motion Generators (pg. 273)
10.3 Learning the Desired Impedance Profiles (pg. 285)
10.4 Passive Interaction Control with DSs (pg. 287)
11. Force Control (pg. 295)
11.1 Motion and Force Generation in Contact Tasks with DSs (pg. 295)
12. Conclusion and Outlook (pg. 303)
V. Appendices (pg. 305)
A. Background on Dynamical Systems Theory (pg. 307)
A.1 Dynamical Systems (pg. 307)
A.2 Visualization of Dynamical Systems (pg. 308)
A.3 Linear and Nonlinear Dynamical Systems (pg. 308)
A.4 Stability Definitions (pg. 309)
A.5 Stability Analysis and Lyapunov Stability (pg. 311)
A.6 Energy Conservation and Passivity (pg. 312)
A.7 Limit Cycles (pg. 313)
A.8 Bifurcations (pg. 314)
B. Background on Machine Learning (pg. 315)
B.1 Machine Learning Problems (pg. 315)
B.2 Metrics (pg. 316)
B.3 Gaussian Mixture Models (pg. 319)
B.4 Support Vector Machines (pg. 337)
B.5 Gaussian Processes Regression (pg. 348)
C. Background on Robot Control (pg. 357)
C.1 Multi-rigid Body Dynamics (pg. 357)
C.2 Motion Control (pg. 357)
D. Proofs and Derivations (pg. 361)
D.1 Proofs and Derivations for Chapter 3 (pg. 361)
D.2 Proofs and Derivations for Chapter 4 (pg. 362)
D.3 Proofs and Derivations for Chapter 5 (pg. 363)
D.4 Proofs and Derivations for Chapter 9 (pg. 373)
Notes (pg. 379)
Bibliography (pg. 383)
Index (pg. 391)

Aude Billard

Aude Billard is Professor, School of Engineering, Ecole Polytechnique Fédérale de Lausanne (EPFL) and Director of the Learning Algorithms and Systems Laboratory (LASA).


Sina Mirrazavi

Sina Mirrazavi is a Senior Researcher at Sony.


Instructors
You must have an instructor account and submit a request to access instructor materials for this book.
eTextbook
Go paperless today! Available online anytime, nothing to download or install.