Modular Design Automation of Intelligent Robot Systems

Zhun Fan, PhD, is a Full Professor at the Shenzhen Institute for Advanced Study, University of Electronic Science and Technology, China.Zhaojun Wang is currently pursuing a PhD degree with the Department of Civil Engineering, College of Engineering at Shantou University.Wenji Li, PhD, is a Lecturer in the Department of Electronics at the College of Engineering, Shantou University.Guijie Zhu is currently pursuing a PhD degree with the Department of Civil Engineering, College of Engineering at Shantou University.Jiafan Zhuang, PhD, joined Shantou University as a lecturer in September of 2022, and currently leads a computer vision research group.

• Acronyms ix1 Introduction to Design Automation and Intelligent Robotic Systems 11.1 Background of Design Automation 11.1.1 Historical Evolution 11.1.2 Importance in Modern Engineering 31.2 Intelligent Robotic Systems 31.2.1 Definition and Components 31.2.2 Challenges in Design and Optimization 41.3 Overview of MODENA 51.4 Objectives and Significance of the Book 62 Foundations of MODENA 92.1 Principles of Modular Design Automation 92.2 Evolutionary Computation in MODENA 112.2.1 Fundamentals 112.2.2 Application in Design Automation 182.3 Neural Architecture Search 292.3.1 Theory and Algorithms 302.3.2 Application in Design Automation 312.4 Causal Discovery 352.5 Comparison with Traditional and LLM-based Methods 382.5.1 Contrast with Traditional Approaches 382.5.2 Differences from LLM-based Methods 393 Optimization Methods in MODENA 413.1 General Framework of PPS 423.2 Pps-moea/d 463.2.1 Basic Description of PPS-MOEA/D 463.2.2 Experimental Study 533.2.3 Conclusion 633.3 Pps-m2m 633.3.1 Basic Description of M2M Strategy 643.3.2 PPS Search Strategy 643.3.3 Combination of PPS with M2M 663.3.4 Key Differences Between Two PPS-based Algorithms 693.3.5 Experiment Results 693.3.6 Conclusion 843.4 Sa-pps 853.4.1 Proposed Method 853.4.2 Push Search Stage 883.4.3 Pull Search Stage 893.4.4 BatchBALDS 953.4.5 Experimental Study 983.4.6 Conclusion 1043.5 Genetic U-Net 1053.5.1 Search Space and Encoding Mechanism 1063.5.2 Evolutionary Algorithm 1103.5.3 Experimental Setup 1173.5.4 Results and Analysis 1193.5.5 Conclusion 1233.6 Evolving Hybrid Bond Graph Using GP 1233.6.1 Bond Graph and GP 1263.6.2 Basic Primitives 1283.6.3 Genetic Operators 1333.6.4 Conclusion 1374 Application of MODENA 1394.1 Morphology Design Automation 1394.1.1 Teaching Manipulator Design 1404.1.2 Optimal Design of Drive Mechanism for Electric Typewriter 1534.2 Controller Design Automation 1644.2.1 Discrete Controller Design for Hybrid Mechatronic Systems 1644.2.2 Evolution Design for Vehicle Suspension System 1854.3 Vision System Design Automation 1974.3.1 Optimization Design for the Retinal Vascular System 1974.3.2 Causal Feature Selection for Strabismus Diagnosis Using GNN 2105 MODENA in Swarm Robotics 2175.1 Introduction 2175.2 Automated Swarm Pattern Generation for Swarm Robots 2185.2.1 Background and Problem Formulation 2185.2.2 Automated GRN Model Structure Design 2215.2.3 Experimental Results 2285.2.4 Conclusion 2405.3 Swarm Control in Communication-denied Environments 2405.3.1 Method Architecture of VG-Swarm 2415.3.2 Experimental Setup 2495.3.3 Experimental Results 2515.3.4 Conclusion 2545.4 Vision-based Distributed Multi-UAV Collision Avoidance 2555.4.1 Related Work 2565.4.2 Proposed Method 2575.4.3 Experiments and Results 2625.4.4 Conclusion 2685.5 Multi-UAV Collision Avoidance via Causal Representation Learning 2685.5.1 Related Work 2685.5.2 Proposed Method 2715.5.3 Experiment Results and Analysis 2745.5.4 Conclusion 2786 Conclusions and Future Directions 2796.1 Conclusions 2796.2 Research Prospects 279References 283Index 319