Intelligent Control Systems with an Introduction to System of Systems Engineering Book

Preface xiii

About the Authors xv

1 Introduction 1

References 7

2 Elements of a Classical Control System 9

2.1 Introduction 9

2.1.1 Plant 9

2.1.2 Variables 10

2.1.3 System 10

2.1.4 Disturbances 11

2.2 How the Model of a Dynamic System Can Help to Control It 11

2.2.1 Newtonian Dynamics 11

2.2.2 Control Experiments on a Point Mass 12

2.2.3 Projects 15

2.3 Control of Robot Manipulators 16

2.3.1 Derivation of Dynamics of a Robot Manipulator 16

2.3.2 Worked Out-Project 1: Manipulator Dynamics 21

2.3.3 Worked Out Project 2: Manipulator Dynamics 23

2.3.4 Effect of Parameter Uncertainty on Model-Based Control Performance 24

2.3.5 Worked Out Project 3: Hybrid Controllers 27

2.4 Stability 31

2.4.1 Equilibrium Points and Their Stability 31

2.4.2 Lyapunov Stability 34

2.4.3 Worked Out Example 35

References 36

3 Introduction to System of Systems 37

3.1 Introduction 37

3.2 Definitions of SoS 38

3.3 Challenging Problems in SoS 39

3.3.1 Theoretical Problems 39

3.3.1.1 Open Systems Approach to SoSE 39

3.3.1.2 Engineering of SoS 40

3.3.1.3 Standards of SoS 40

3.3.1.4 SoS Architecting 41

3.3.1.5 SoS Simulation 41

3.3.1.6 SoS Integration 42

3.3.1.7 Emergence in SoS 43

3.3.1.8 SoS Management: Governance of Paradox 44

3.3.2 Implementation Problems for SoS 45

3.3.2.1 SE for the Department of Defense SoS 45

3.3.2.2 E-enabling and SoS Aircraft Design via SoSE 46

3.3.2.3 An SoS Perspective on Infrastructures 47

3.3.2.4 Sensor Networks 48

3.3.2.5 An SoS View of Services 48

3.3.2.6 SoSE in Space Exploration 50

3.3.2.7 Communication and Navigation in Space SoS 50

3.3.2.8 Electric Power Systems Grids as SoS52

3.3.2.9 SoS Approach for Renewable Energy 52

3.3.2.10 Sustainable Environmental Management from an SoSE Perspective 53

3.3.2.11 Robotic Swarms as an SoS 54

3.3.2.12 Transportation Systems 55

3.3.2.13 Health Care Systems 57

3.3.2.14 Global Earth Observation SoS 58

3.3.2.15 Deepwater Coastguard Program 59

3.3.2.16 Future Combat Missions 60

3.3.2.17 National Security 61

3.4 Conclusions 62

References 62

4 Observer Design and Kalman Filtering 67

4.1 State Space Methods for Model-Based Control 67

4.1.1 Derivation of Dynamics for the Cart-Pole Balancing Problem 67

4.1.2 Introduction to State Space Representation of Dynamics 70

4.1.3 Regulator Control of State Vectors 72

4.2 Observing and Filtering Based on Dynamic Models 73

4.2.1 Why Observers? 74

4.2.2 Elements of an Observer 74

4.2.3 Simulations on an Inverted Pendulum Control 77

4.3 Derivation of the Discrete Kalman Filter 83

4.4 Worked Out Project on the Inverted Pendulum 87

4.5 Particle Filters 94

References 97

5 Fuzzy Systems-Sets, Logic, and Control 99

5.1 Introduction 99

5.2 Classical Sets 101

5.3 Classical Set Operations.102

5.3.1 Union 102

5.3.2 Intersection 102

5.3.3 Complement 103

5.4 Properties of Classical Set 103

5.5 Fuzzy Sets 104

5.6 Fuzzy Set Operations 106

5.6.1 Union 106

5.6.2 Intersection 108

5.6.3 Complement 108

5.7 Properties of Fuzzy Sets 108

5.7.1 Alpha-Cut Fuzzy Sets 109

5.7.1.1 Alpha-Cut Sets 110

5.7.2 Extension Principle 110

5.8 Classical Relations versus Fuzzy Relations 112

5.9 Predicate Logic 114

5.9.1 Tautologies 118

5.9.2 Contradictions 118

5.9.3 Deductive Inferences 119

5.10 Fuzzy Logic 121

5.11 Approximate Reasoning 123

5.12 Fuzzy Control 124

5.12.1 Inference Engine 126

5.12.2 Denazification 129

5.12.3 Fuzzy Control Design 129

5.12.4 Analysis of Fuzzy Control Systems 132

5.12.5 Stability of Fuzzy Control Systems 135

5.12.5.1 Time-Domain Methods 136

5.12.5.2 Frequency-Domain Methods 138

5.12.6 Lyapunov Stability 138

5.12.7 Stability via Interval Matrix Method 143

5.13 Conclusions 147

References 147

6 Neural Network-Based Control 151

6.1 Introduction to Function Approximation 151

6.1.1 Biological Inspirations 151

6.1.2 Construction of Complex Functions by Adaptive Combination of Primitives 160

6.1.3 Concept of Radial Basis Functions 161

6.2 NN-Based Identification of Dynamics of a Robot Manipulator 165

6.2.1 An Approach to Estimate the Elements of the Mass, Centrifugal, Coriolis, and Gravity Matrices 168

6.2.2 Optimum Parameter Estimation 169

6.3 Structure of NNs 174

6.4 Generating Training Data for an NN 178

6.5 Dynamic Neurons 179

6.6 Attractors, Strange Attractors, and Chaotic Neurons 181

6.6.1 Attractors and Recurrent Neurons 181

6.6.2 A Chaotic Neuron 183

6.7 Cerebellar Networks and Exposition of Neural Organization to Adaptively Enact Behavior 184

References 188

7 System of Systems Simulation 189

7.1 Introduction 189

7.2 SoS in a Nutshell 190

7.3 An SoS Simulation Framework 195

7.3.1 DEVS Modeling and Simulation 195

7.3.2 XML and DEVS 196

7.4 SoS Simulation Framework Examples 197

7.4.1 Case 1: Data Aggregation Simulation 198

7.4.1.1 DEVS-XML Format 198

7.4.1.2 Programming Environment 199

7.4.1.3 Simulation Results 200

7.4.2 Case Study 2: A Robust Threat Detection System Simulation 202

7.4.2.1 DEVS-XML Format 202

7.4.2.2 Simulation Setup 203

7.4.2.3 Robust Threat Detection Simulation 205

7.4.3 Case 3: Threat Detection Scenario with Several Swarm Robots 208

7.4.3.1 XML Messages 209

7.4.3.2 DEVS Components of the Scenario 210

7.4.3.3 DEVS-XML SoS Simulation 215

7.5 Agent-in-the-Loop Simulation of an SoS 222

7.5.1 XML SoS Real-Time Simulation Framework 223

7.5.1.1 Threat Detection Scenario 223

7.5.1.2 Synchronization 223

7.5.2 DEVS Modeling 224

7.5.2.1 DEVS Components 224

7.5.2.2 Mobile Swarm Agent 224

7.5.2.3 Base Station 226

7.5.3 Agent-in-the-Loop Simulation 227

7.6 Conclusion 228

Acknowledgment 228

References 229

8 Control of System of Systems 233

8.1 Introduction 233

8.2 Hierarchical Control of SoS 233

8.3 Decentralized Control of SoS 240

8.3.1 Decentralized Navigation Control 243

8.3.2 The Decentralized Control Law 244

8.3.2.1 Motion Coordination 245

8.4 Other Control Approaches 252

8.4.1 Consensus-Based Control 252

8.4.2 Cooperative Control 255

8.4.3 Networked Control 255

8.5 Conclusions 258

References 259

9 Reward-Based Behavior Adaptation 261

9.1 Introduction 261

9.1.1 Embodiment 263

9.1.2 Situatedness 263

9.1.3 Internal Models 264

9.1.4 Policy 264

9.1.5 Reward 264

9.1.6 Emergence 264

9.2 Markov Decision Process 265

9.2.1 A Markov State 265

9.2.2 Value Function 266

9.3 Temporal Difference-Based Learning 267

9.4 Extension to Q Learning 270

9.5 Exploration versus Exploitation 271

9.6 Vector Q Learning 272

9.6.1 Summary of Results 276

References 279

10 An Automated System to Induce and Innovate Advanced Skills in a Group of Networked Machine Operators 281

10.1 Introduction 281

10.2 Visual Inspection and Acquisition of Novel Motor Skills 282

10.3 Experimental Setup 283

10.4 Dynamics of Successive Improvement of Individual Skills 285

10.5 Proposed Model of Internal Model Construction and Learning 289

10.6 Discussion and Conclusion 294

References 294

11 A System of Intelligent Robots-Trained Animals-Humans in a Humanitarian Demining Application 297

11.1 Introduction 297

11.1.1 Mine Detection Technology 298

11.1.2 Metal Detectors (Electromagnetic Induction Devices) 299

11.1.3 Ground-Penetrating Radar 299

11.1.4 Multisensor Systems Using GPR and Metal Detectors 300

11.1.5 Trace Explosive Detection Systems 301

11.1.6 Biosensors 301

11.1.7 Magnetic Quadrupole Resonance 301

11.1.8 Seismoacoustic Methods 301

11.2 A Novel Legged Field Robot for Landmine Detection 302

11.2.1 Key Concerns Addressed by the Moratuwa University Robot for Anti-Landmine Intelligence Design 302

11.2.2 Key Design Features of MURALI Robot 303

11.2.3 Designing the Robot with Solidworks 305

11.2.4 Motherboard to Control the Robot 333

11.2.5 Basics of Sensing and Perception 336

11.2.6 Perception of Environment and Intelligent Path Planning 339

11.3 Combining a Trained Animal with the Robot 346

11.3.1 Basic Background on Animal-Robot interaction 347

11.3.2 Background on Reward-Based Learning 348

11.3.3 Experience with Training a Mongoose 348

11.3.3.1 Phase 1: Conditioning Smell, Reward, and Sound 348

11.3.3.2 Phase 2: Learning in a Paradigm Where the Degree of Difficulty of Correct Classification Was Progressively Increased 349

11.4 Simulations on Multirobot Approaches to Landmine Detection 353

References 359

12 Robotic Swarms for Mine Detection System of Systems Approach 363

12.1 Introduction 363

12.1.1 Swarm Intelligence 363

12.1.2 Robotic Swarms 364

12.1.3 System-of Systems 365

12.2 SoS Approach to Robotic Swarms366

12.2.1 Interoperability 366

12.2.2 Integration 367

12.3 Designing System of Swarm Robots: GroundScouts 368

12.3.1 Hardware Modularity 369

12.3.1.1 Locomotion 370

12.3.1.2 Control 371

12.3.1.3 Sensor 372

12.3.1.4 Communication 373

12.3.1.5 Actuation 373

12.3.2 Software Modularity 374

12.3.2.1 Operating System 375

12.3.2.2 Dynamic Task Uploading 375

12.3.3 Communication Protocol: Adaptive and Robust 376

12.3.3.1 Physical Layer 377

12.3.3.2 MAC Layer 377

12.3.3.3 Implementation Results 381

12.4 Mine Detection with Ant Colony-Based Swarm Intelligence 382

12.4.1 Simulation of Mine Detection with ACO 383

12.4.1.1 Arit Colony Optimization 383

12.4.1.2 The Mine Detection Problem 385

12.4.1.3 ACO Model Used for Mine Detection 387

12.4.2 Implementation of ACO-Based Mine Detection Algorithm on GroundScouts 397

12.4.2.1 Implementation of Mines 398

12.4.2.2 Implementation of the Scent 399

12.4.2.3 Main Program 400

12.4.2.4 Graphical User Interface 403

12.4.3 Experimental Results 406

12.5 Conclusion 409

Acknowledgment 410

References 410

Index 417