Autonomous Marine Vehicles Planning and Control

Yong Bai, PhD is a professor in the College of Civil Engineering and Architecture at Zhejiang University. He has written over 200 academic papers in national and international academic journals and internationally published over 20 books. His research interests include marine engineering structures, unmanned surface vehicles, autonomous underwater vehicles, hydrogen vessels, marine pipelines and risers, engineering risk analysis, and safety assessment. Liang Zhao, PhD is a research fellow at Zhejiang University. He has co-authored over 20 research articles in top engineering journals. His current research focuses on planning and decision making for marine robotics, asynchronous maritime perception, and green and intelligent shipping.

• Preface v1 Introduction 11.1 Overview 11.2 System Structure 61.3 Mathematical Model of a USV 81.4 Maritime Applications 111.5 Motivation of this Book 13References 132 Automatic Control Module 152.1 Origin and Development 162.2 Common Control System Development 172.2.1 Dynamic Positioning and Position Mooring Systems 172.2.1.1 Dynamic Positioning Control System 182.2.1.2 Position Mooring Control System 222.2.2 Waypoint Tracking and Path-Following Control Systems 242.2.2.1 Waypoint Tracking Control System 242.2.2.2 Path-Following Control System 262.3 Advanced Control System Development 312.3.1 Linear Quadratic Optimal Control 312.3.2 State Feedback Linearization 362.3.2.1 Decoupling in the BODY Frame (Velocity Control) 362.3.2.2 Decoupling in the NED Frame (Position and Attitude Control) 382.3.3 Integrator Backstepping Control 402.3.4 Sliding-Mode Control 452.3.4.1 SISO Sliding-Mode Control 452.3.4.2 Sliding-Mode Control Using the Eigenvalue Decomposition 49References 523 Perception and Sensing Module 573.1 Low-Pass and Notch Filtering 583.1.1 Low-Pass Filtering 583.1.2 Cascaded Low-Pass and Notch Filtering 593.2 Fixed Gain Observer Design 603.2.1 Observability 603.2.2 Luenberger Observer 603.2.3 Case Study: Luenberger Observer for Heading Autopilots Using Only Compass Measurements 613.3 Kalman Filter Design 613.3.1 Discrete-Time Kalman Filter 613.3.2 Continuous-Time Kalman Filter 623.3.3 Extended Kalman Filter 633.3.4 Corrector–Predictor Representation for Nonlinear Observers 643.3.5 Case Study: Kalman Filter for Heading Autopilots Using Only Compass Measurements 643.3.5.1 Heading Sensors Overview 643.3.5.2 System Model for Heading Autopilot Observer Design 653.3.6 Case Study: Kalman Filter for Dynamic Positioning Systems Using GNSS and Compass Measurements 663.4 Nonlinear Passive Observer Designs 673.4.1 Case Study: Passive Observer for Dynamic Positioning Using GNSS and Compass Measurements 673.4.2 Case Study: Passive Observer for Heading Autopilots Using only Compass Measurements 683.4.3 Case Study: Passive Observer for Heading Autopilots Using Both Compass and Rate Measurements 713.5 Integration Filters for IMU and Global Navigation Satellite Systems 713.5.1 Integration Filter for Position and Linear Velocity 723.5.2 Accelerometer and Compass Aided Attitude Observer 733.5.3 Attitude Observer Using Gravitational and Magnetic Field Directions 73References 744 Model Predictive Control for Autonomous Marine Vehicles: A Review 754.1 Introduction 754.1.1 Object Introduction 754.1.2 Previous Reviews 774.2 Fundamental Models and a General Picture 854.2.1 Model of AMVs 854.2.1.1 6-DOF Model 854.2.1.2 3-DOF Model 904.2.2 Model Predictive Control 924.2.3 Literature Search 964.3 Methodology 994.3.1 MPC Applications of AMVs 994.3.1.1 Real-Coded Chromosome 994.3.1.2 Path Following 1014.3.1.3 Trajectory Tracking 1044.3.1.4 Cooperative Control/Formation Control 1064.3.1.5 Collision Avoidance 1084.3.1.6 Energy Management 1114.3.1.7 Other Topics 1134.4 Discussion 1144.4.1 Limitations of Existing Techniques and Challenges in Developing MPC 1144.4.1.1 Uncertainties of AMV Motion Models 1144.4.1.2 Stability and Security of the New MPC Method 1154.4.1.3 The Balance Between Effectiveness and Efficiency of the Methods 1154.4.1.4 The Practical Application Scenario of the MPC and the Discussion of the Working Conditions 1164.4.1.5 Challenges Posed by the Marine Environment Affect MPC Development for AMVs 1164.4.2 Trends in the Technology Development for MPC in AMV 1174.4.2.1 More Cooperative Control with MPC 1174.4.2.2 Rigorous Theoretical Derivation and Experimental Verification 1174.4.2.3 Real-Time MPC for AMVs Applications 1184.4.2.4 The Combination of Machine Learning/Neural Networks and MPC for AMVs Applications 1184.4.2.5 Address the Challenges Posed by the Marine Environment 1194.4.2.6 Potential Interdisciplinary Approaches that Combine MPC with Other Innovative Fields 1204.5 Conclusion 121Acknowledgement 121References 1215 Controller-Consistent Path Planning for Unmanned Surface Vehicles 1295.1 Introduction 1295.2 Problem Formulation 1315.3 Methodology 1325.3.1 Improved Artificial Fish Swarm Algorithm 1325.3.1.1 Prey Behavior 1335.3.1.2 Follow Behavior 1355.3.1.3 Swarm Behavior 1355.3.1.4 Random Behavior 1365.3.1.5 Adaptive Visual and Step 1365.3.2 Expanding Technique 1385.3.3 Node Cutting and Path Smoother 1395.3.4 Establishment of USV Model 1415.4 Simulation 1445.4.1 Monte Carlo Simulation 1455.4.2 Path Quality Test 1465.4.3 Simulation Using USV Control Model in Practical Environment 1495.5 Conclusion 151References 1526 Nonlinear Model Predictive Control and Routing for USV-Assisted Water Monitoring 1556.1 Introduction 1566.2 Problem Formulation 1616.2.1 Heterogeneous Global Path Planning Problem 1616.2.1.1 USV Model 1616.2.1.2 Task Model 1626.2.1.3 Problem Statement 1626.2.2 Problem Analysis 1646.2.3 Path Following Problem 1646.2.3.1 Basic Assumptions 1656.2.3.2 Vessel Model 1656.2.3.3 Problem Description 1686.3 Methodology 1696.3.1 Greedy Partheno Genetic Algorithm 1696.3.1.1 Dual-Coded Chromosome 1706.3.1.2 Fitness Function 1706.3.1.3 Greedy Randomized Initialization 1716.3.1.4 Local Exploration 1726.3.1.5 Mutation Operators 1746.3.1.6 Algorithm Flow 1756.3.2 Nonlinear Model Predictive Control 1776.3.2.1 State Space Model 1776.3.2.2 NMPC Design 1786.3.2.3 Solver 1806.3.2.4 Stability 1816.4 Results and Discussion 1816.4.1 Simulation: Global Task Planning 1816.4.1.1 Convergence Test 1816.4.1.2 Heterogeneous Task Planning 1856.4.2 Simulation: NMPC Control Performance 1886.4.2.1 Test 1: Simulation Under Different Model Uncertainties 1906.4.2.2 Test 2: Comparative Study with Other Methods 1926.4.3 Simulation Verification of the Framework 1966.5 Conclusion 200References 2017 Global-Local Hierarchical Framework for USV Trajectory Planning 2077.1 Introduction 2077.2 Problem Formulation 2127.2.1 Marine Environment 2127.2.2 Dynamic Obstacles 2137.2.3 Effects of Currents 2137.2.4 USV Model and Constraints 2137.2.5 Protocol Constraints 2167.2.6 Objective Functions 2177.2.6.1 The Minimum Cruising Time 2177.2.6.2 The Minimum Variation of Heading Angle 2177.2.6.3 The Safest Path 2187.2.7 Problem Statement 2197.3 Methodology 2217.3.1 Adaptive-Elite GA with Fuzzy Inference (AEGAfi) 2217.3.1.1 Real-Coded Chromosome 2217.3.1.2 Initialization Based on Adaptive Random Testing (ART) 2227.3.1.3 Adaptive Elite Selection 2237.3.1.4 Double-Functioned Crossover 2247.3.1.5 Mutation Operators 2257.3.1.6 Fuzzy-Based Probability Choice 2267.3.1.7 Fitness Function Design 2277.3.2 Replanning Strategy Based on Sensory Vector 2297.3.2.1 Sensory Vector Structure 2297.3.2.2 Formulation of V s 2307.3.2.3 Formulation of Gap Vector V g Based on COLREGs 2327.3.2.4 Formulation of Transition Path 2347.4 Simulation Study 2367.4.1 Convergence Benchmark Analysis 2367.4.2 Simulation Under Static Environment 2387.4.3 Simulation Under Time-Varying Environment 2467.4.4 Simulation on Real-World Geography 2517.5 Conclusion 254Appendix 255List of Abbreviations 255Acknowledgements 256References 2568 Reinforcement Learning for USV-Assisted Wireless Data Harvesting 2638.1 Introduction 2638.2 Fundamental Models 2698.2.1 Environment Model 2728.2.2 Sensor Node and Communication Model 2738.2.3 USV Model 2758.2.3.1 Kinematic Model 2758.2.3.2 Sensing Module 2778.3 Methodology 2788.3.1 Brief States on Q-Learning 2788.3.2 Interactive Learning 2798.3.2.1 Heuristic Reward Design 2798.3.2.2 Design of Value-Iterated Global Cost Matrix 2798.3.2.3 Local Cost Matrix and Path Generation 2828.3.2.4 USV Actions with Discrete Precise Clothoid Path 2838.3.3 Summary of the Path Planning Algorithm 2868.3.4 Time Complexity 2878.4 Results and Discussion 2888.4.1 Performance Indicators 2888.4.2 Hyper-Parameter Analysis 2908.4.3 Comparative Study with State of the Art 2948.5 Conclusion 298Appendix 299References 3009 Achieving Optimal Dynamic Path Planning for Unmanned Surface Vehicles: A Rational Multi-Objective Approach and a Sensory-Vector Re-Planner 3079.1 Introduction 3089.2 Problem Formulation 3149.2.1 Environment Modeling 3159.2.1.1 Motion Area 3159.2.1.2 Effects of Currents 3159.2.2 Dynamic Obstacles 3169.2.3 Motion Constraints 3179.2.4 Objective Functions 3179.2.4.1 Path Length 3179.2.4.2 Path Smoothness 3189.2.4.3 Energy Consumption 3189.2.4.4 The Safest Path 3189.2.5 Optimization Problem Statement 3199.3 Methodology 3219.3.1 Framework of NSGA-II 3219.3.2 Aensga-ii 3229.3.2.1 Real-Coded Representation 3229.3.2.2 Initialization Using Candidate Set Adaptive Random Testing (CSART) 3239.3.2.3 Adaptive Crowding Distance (ACD) Strategy 3249.3.2.4 Improved Binary Tournament Selection 3269.3.3 Fuzzy Satisfactory Degree 3279.3.4 Replanning Strategy Based on Sensory Vector 3309.3.4.1 Sensory Vector Structure 3309.3.4.2 Formulation of Gap Vector V g Based on COLREGs 3339.3.4.3 Formulation of Transition Path 3359.4 Results and Discussion 3369.4.1 Convergence and Diversity Analysis 3369.4.2 Implementation in Static Environment 3429.4.2.1 Fixed Currents 3429.4.2.2 Time-Varying Currents 3469.4.3 Simulation Under Dynamic Environment 3519.5 Conclusion 356Acknowledgements 357References 35710 Coordinated Trajectory Planning for Multiple AUVs 36310.1 Introduction 36310.1.1 Background 36310.1.2 Related Work 36410.1.3 Contributions 36610.2 Problem Model 36710.2.1 Environment Model 36710.2.2 AUV Model 36910.2.3 Space and Time Constraint Model 37010.2.4 Optimization Terms 37110.2.5 Problem Statement 37410.3 Solver Design 37410.3.1 Brief States on Grey Wolf Optimizer 37410.3.2 Parallel Grey Wolf Optimizer Design 37610.4 Results and Discussion 37910.4.1 Simulation 1: Allocation Task 38010.4.2 Simulation 2: Rendezvous Task 38110.5 Conclusion 385Acknowledgements 385References 38611 Coverage Strategy for USV-Assisted Coastal Bathymetric Mapping 38911.1 Introduction 39011.2 Fundamental Models 39411.2.1 Region of Interest 39411.2.2 USV Model 39511.3 Methodology 39611.3.1 Coastal Line Approximation 39611.3.2 Coverage Strategy 39711.3.2.1 Trapezoidal Cellular Decomposition 39711.3.2.2 Optimal Back and Forth Coverage Algorithm 39811.3.2.3 Theoretical Analysis 40211.3.3 Fuzzy-Biased Random Key Evolutionary Algorithm (FRKEA) 40311.3.3.1 Chromosome Mapping 40411.3.3.2 Evaluation in Real Space 40511.3.3.3 Elitist Breeding 40611.3.3.4 Mutating 40711.3.3.5 Fuzzy Bias 40911.4 Results and Discussion 41111.4.1 Convergence Analysis 41211.4.2 Simulation Study 41411.4.2.1 Competitive Study 41411.4.2.2 Parameter Analysis 41711.4.3 Lake Trials 41911.5 Conclusion 423References 42412 Energy-Efficient Coverage for USV-Assisted Bathymetric Survey Under Currents 42912.1 Introduction 42912.2 Methodology 43312.2.1 Problem Models 43312.2.1.1 Region of Interest 43312.2.1.2 Current Model 43312.2.1.3 USV Kinematics Under Currents 43412.2.1.4 Energy Estimation 43512.2.2 Coverage Strategy 43612.3 Results and Discussion 44012.3.1 Preparation 44012.3.2 Analysis on Polygon Shapes 44112.3.3 Analysis on Attacking Angle 44412.3.4 Analysis on Different Coverage Strategy 44512.3.5 Test on a Complex Concave ROI 44712.4 Conclusion 454References 45513 Modeling and Solving Time-Sensitive Task Allocation for USVs with Mixed Capabilities 45913.1 Introduction 45913.2 Problem Formulation 46313.2.1 Fundamental Models 46313.2.1.1 USV Model 46313.2.1.2 Target Model 46413.2.2 Extended-Restriction Multiple Traveling Salesman Problem (ER-MTSP) 46513.2.3 Problem Analysis 46713.3 Methodology 46813.3.1 Dual-Coded Chromosome Representation 46813.3.2 Adaptive Random Testing Initialization 46913.3.3 Hierarchical Crossover 46913.3.4 Customized Mutation Strategy 47213.3.5 Two-Phase Refinement Strategy 47313.3.6 Linguistic Satisfactory Degree 47513.4 Results and Discussion 47713.4.1 Convergence and Diversity Analysis 47713.4.2 Case Studies 48013.4.3 Field Test 48713.5 Conclusion 492References 49314 Joint Optimized Coverage Planning Framework for USV-Assisted Offshore Bathymetric Mapping: From Theory to Practice 49714.1 Introduction 49814.2 Problem Formulation 50214.2.1 Definitions 50214.2.2 Problem Statement 50314.2.3 Theoretical Analysis 50614.3 Methods for Problem Solving 50714.3.1 Bisection-Based Convex Decomposition 50714.3.2 Hierarchical Heuristic Optimization Algorithm 51014.3.2.1 Order Generation 51014.3.2.2 Candidate Pattern Finding 51414.3.2.3 Tour Finding 51814.3.2.4 Final Optimization 51914.4 Results and Discussion 52014.4.1 Validation in Simulation 52014.4.2 Lake Experiments 52614.5 Conclusion 530Acknowledgements 530Appendix 530References 53015 Pipe Segmentation and Geometric Reconstruction from Poorly Scanned Point Clouds Based on Deep Learning and BIM-Generated Data Alignment Strategies 53515.1 Introduction 53515.2 Related Studies 53715.2.1 Pipe Segmentation 53715.2.1.1 Descriptor-Based Methods 53715.2.1.2 Learning-Based Methods 53815.2.2 Dataset Preparation 53815.2.3 Pipe Reconstruction 53915.3 Methodology 53915.3.1 BIM-Based Data Generating 54015.3.2 Network Architecture 54215.3.2.1 Overall Architecture 54215.3.2.2 PipeSegNet Architecture 54315.3.2.3 Feature Alignment Module 54515.3.2.4. Label Alignment Module 54615.3.2.5 Loss Function 54715.3.3 Pipe Geometric Reconstruction 54815.4 Experiment 55215.4.1 Experimental Settings 55215.4.2 Evaluation Metrics 55515.4.3 Results and Discussion 55615.5 Conclusion 563Acknowledgment 564References 56416 The Arc Routing Path Planning Problem in the Maritime Domain 57116.1 Introduction 57116.2 The Arc Routing Path Planning Problem 57516.2.1 Introduction to Arc Routing 57516.2.2 Common Applications of Arc Routing 57716.3 One Solution for Arc Problem: The Chinese Postman Problem 57816.3.1 Basic Conception 57816.3.2 Core Formulation 57916.3.3 Variants of the Chinese Postman Problem 58016.3.4 Algorithmic Approaches and Solution Methods 58116.3.4.1 Polynomial-Time Solutions 58116.3.4.2 NP-Hard Variants 58216.4 Case Study 58316.4.1 Background 58316.4.2 Platform Design 58416.4.3 Full Coverage Problem 58616.4.3.1 Theoretical Formulation: Using the Chinese Postman Problem for Efficient Coverage 58616.4.3.2 Coverage Path Generation 58716.4.3.3 Discussion 58816.5 Concluding Remarks 588References 58917 Atmospheric Scattering Model-Based Dataset for Maritime Object Detection with YOLOv 11 59117.1 Introduction 59117.2 Methodology 59317.2.1 Physics-Based Fog Simulation Using Depth Estimation 59317.2.1.1 MiDaS: Monocular Depth Estimation 59317.2.1.2 Atmospheric Scattering Model 59517.2.2 YOLOv 11 59617.3 Experiment 59817.3.1 Dataset 59817.3.2 Foggy Dataset Generation and Model Training 59917.3.2.1 Foggy Dataset Generation 59917.3.2.2 Model Training 59917.4 Result and Discussion 60017.4.1 Baseline Training and Generalization Analysis 60017.4.2 Improving Model Robustness with Mixed- Concentration Fog Training 60117.4.3 Detection Result Comparison 60417.5 Conclusion 610References 61118 Multisensor Perception and Data Fusion Technologies 61318.1 Camera-Based Detection Approaches 61418.1.1 RGB and Stereo Camera 61418.1.2 Infrared and Thermal Camera 61718.1.3 Object Detection Methodologies 61818.2 LiDAR-Based Detection Approaches 62018.2.1 Stages of Object Detection 62118.2.2 Challenges and Resolutions 62318.3 Data Fusion Methods 62418.3.1 Radar 62518.3.2 Fusion Level 62618.3.3 Synchronization and Calibration 627References 62919 Route Planning for Low-Altitude UAV Using Multi-Objective Optimization 63319.1 Introduction 63419.2 Problem Model 63619.3 Multi-Objective Particle Swarm Optimization 63919.4 Results and Discussion 643References 64520 Autonomous System Design of Marine Vehicles 64720.1 Introduction 64720.2 Planning Module Design 64920.2.1 Recursive Cell Decomposition Method 65020.2.2 Optimal Path Generation 65320.2.3 Guidance Planning: Adaptive Line-of-Sight (ALOS) Method 65620.3 Control Module Design: USV Dynamics Modeling 65720.4 Combined Navigation Module Design 661References 663Index 665