Title Page
Abstract
Contents
Chapter 1. Introduction 29
1.1. Motivations 29
1.2. Contributions 32
1.3. Thesis Outline 34
Chapter 2. Automated Sewing System Enabled by Machine Vision for Smart Garment Manufacturing 36
2.1. Background 36
2.2. Algorithms 40
2.2.1. Instance segmentation based on YOLOv5 40
2.2.2. Seam line detection and path generation 45
2.3. System 50
2.3.1. Automatic sewing machine 51
2.3.2. Communication 52
2.4. Experiments and results 53
2.4.1. Instance segmentation 53
2.4.2. Seam line detection and top stitch path generation 55
2.4.3. Top stitching with automated sewing machine 57
2.5. Demonstration 60
2.6. Discussion 61
Chapter 3. Part I. A Microneedle-Assisted Soft Gripper for Delicate Robotic Manipulation of Fabric 63
3.1. Backgrounds 63
3.2. Design 66
3.2.1. Gripper design 66
3.2.2. Analysis 68
3.2.3. Operation procedure 70
3.2.4. Fabrication 70
3.3. Experiments and results 71
3.3.1. Model evaluation 73
3.3.2. Comparison test 75
3.3.3. Durability test 77
3.3.4. Single-sheet handling 78
3.4. Demonstration 80
3.4.1. Task performs based on the number of gripped sheets 80
3.4.2. Handling high air permeability fabric 81
3.4.3. Automatic stamping device 82
3.4.4. Fabric handling with mobile platform 83
3.5. Discussion 84
Chapter 3. Part II. Multi-Functional Soft Gripper with Microneedles and Integrated Sensing for Robotic Fabric Handling 86
3.6. Backgrounds 86
3.7. Design 90
3.7.1. Operation procedure 90
3.7.2. Integrated pressure sensor and multi-functional structure 91
3.7.3. Soft gripper with multi-modes actuation 92
3.7.4. Analysis 93
3.7.5. Simulations 96
3.7.6. Fabrication 100
3.8. Experiments and results 101
3.8.1. Multi-modes actuation in single structure 103
3.8.2. Pressure response 105
3.8.3. Single sheet separation 109
3.8.4. Compliant structure 111
3.8.5. Characterization of pressure sensor 114
3.9. Demonstration 115
3.10. Discussion 117
Chapter 4. Soft Modularized Robotic Arm for Safe Human-Robot Interaction Based on Visual and Proprioceptive Feedback 120
4.1. Backgrounds 120
4.2. Design 125
4.2.1. 3D-printed tribo-sensitive soft bellow 125
4.2.2. Modular design 127
4.2.3. Material selection 128
4.3. Fabrication 130
4.3.1. Tribo-sensitive soft bellow 130
4.3.2. Soft string strain sensor 131
4.4. Modeling 133
4.4.1. Forward kinematics 133
4.4.2. Inverse kinematics 136
4.5. Simulation 139
4.5.1. Linear translation 139
4.5.2. Bending 140
4.6. Manipulator module localization 142
4.6.1. String strain sensor-based localization 142
4.6.2. Vision-based localization 148
4.7. Experiment 152
4.7.1. Pressure response 152
4.7.2. Contact recognition 158
4.7.3. Control of manipulator module 161
4.8. Demonstration 168
4.8.1. Interaction based task performs 168
4.8.2. Robotic manipulation of fabric using soft multi-functional gripper 170
4.9. Discussion 171
Chapter 5. Conclusion 173
Bibliography 176
국문초록 190
Table 3.1. Information on the fabrics used in model calculation 70
Table 3.2. Information on the fabrics used in the experiments 71
Table 3.3. Results of single-sheet handling test with the numbers of successful trials (success rate) 79
Table 3.4. Information on the fabrics used in experiments 100
Table 4.1. Calculation of LOF means and standard deviations according to the pressure and the number of samples 161
Fig. 2.1. (a) Description of stitching and (b) its actual image. (c) Main components of an automated sewing machine. (d) Acrylic template... 40
Fig. 2.2. Simplified architecture of YOLOv5. (a) Backbone for extracting features. (b) Neck for mixing features. (c) Head for... 41
Fig. 2.3. Vision setup for taking images of templates. 42
Fig. 2.4. Fabrics used in experiments. (a) Black, (b) orange, and (c) blue colored fabric, and their magnified images near the seam line.... 43
Fig. 2.5. (a) Procedures of preprocessing. i. Resizing. ii. Applying contrast-limited adaptive histogram equalization and Gaussian blur. iii.... 45
Fig. 2.6. Schematic of seam line detection principle and pixel intensity plots in different d and p locations. 46
Fig. 2.7. Components of automated sewing machine. 52
Fig. 2.8. Schematic diagram of communication in the automated sewing system. 53
Fig. 2.9. Sequence and result of seam line detection and top stitch path generation. Black fabric with exposure time of (a) 30,000 and (b)... 54
Fig. 2.10. Results of the seam line detection (a) embossed, (b) weaved pattern and curved, (c) and (d) diagonal. (e) White paper insertion for... 56
Fig. 2.11. Robustness test for the proposed algorithm. (a) A single strand of the edge thread is unwound (Left) and a large dirt is place... 57
Fig. 2.12. Automatic sewing machine basic performances. The goal and the current measured position by the absolute encoder in (a) x-... 59
Fig. 2.13. Top stitching result using the custom-built sewing machine. (a) Fabric with marked line. (b) Detection result. (c) Top stitching... 60
Fig. 2.14. Automation procedure for stitching. (a) Left and right images of template (top) and stitched image with detected seam line... 61
Fig. 3.1. (a) Oral disc with sharp teeth of Pacific lamprey(Lampetratridentata). (b) Top view of microneedle-embedded soft... 66
Fig. 3.2. (a) Schematic of the proposed gripping mechanism. As and Af are the areas of suction (2t × w') and pinching ((6 − 2t)/2 × w'),...[이미지참조] 68
Fig. 3.3. Magnified photos of the tested fabrics: (a) Type 1-1: Double-mesh French terry, (b) Type 1-2: Coolmax, (c) Type 2-1: Canvas, (d)... 72
Fig. 3.4. Schematic of the experimental setup 73
Fig. 3.5. (a) Pressure response of the vacuum pump. (b) Pressure response of the gripper. (c) Vertical force measured when the gripper... 74
Fig. 3.6. (a) Performance comparison between grippers with and without needles: pressure difference between the inside and the... 77
Fig. 3.7. Results of durability test: (a) change in the pressure difference during a single cycle of gripping and releasing with △P... 78
Fig. 3.8. (a) Capacitance between parallel electrodes. (b) Schematic and actual prototype of electrodes-integrated gripper. (c) Result of... 81
Fig. 3.9. (a) Fabric with high air permeability. (b) Modified gripper prototype holding the fabric shown in (a). (c) Magnified image of the... 82
Fig. 3.10. (a) Prototype of pattern-numbering machine composed of soft gripper, three degrees-of-freedom robotic arm, and stamp for... 83
Fig. 3.11. Developed mobile platforms. Version 1 for left and Version 2 for right. 84
Fig. 3.12. (a) Photo and (b) schematic front view of the proposed gripper showing the main components. (c) Magnified views of the... 90
Fig. 3.13. (a)-(d) Gripper operation procedure corresponding to combined suction and pinching working principles. (e) Overall design... 93
Fig. 3.14. (a) Suction mode and free body diagram. (b) Pinching mode and free body diagram. (c) Closed pneumatic system schematic... 96
Fig. 3.15. Simulation results of normal force in pinching. Tensile test setup (a) and result (b). (c) Membrane shapes for simulation. (d)... 99
Fig. 3.16. Fabrication process. (a) 3D printing of the gripper body. (b) Microneedles and its holder. (c) Final assembly. 101
Fig. 3.17. (a) - (g), air-permeable fabrics. (h) - (l), non-air-permeable fabrics. (i) - (l) indicate that front and back of the fabrics. 102
Fig. 3.18. Experimental setup to determine characteristics of the gripper. 103
Fig. 3.19. (a) Operational sequence of the two actuation modes. (b) Comparison of the fabric surfaces before and after gripping showing... 105
Fig. 3.20. Experimental results. (a) Pressure responses of suction and pinching. (b) Hysteresis during pinching. (c) Holding force... 108
Fig. 3.21. Simplified schematics of the gripper tip and microneedles. (a) Without microneedles. (b) With microneedles completely embedded... 109
Fig. 3.22. (a) Force range to separate single sheet of air-permeable fabric. (b) Gripping results with pressing force of i. 1 N, ii. 1.5 N, and... 110
Fig. 3.23. Single sheet separation result. Pinching and suction modes are used for Type 1 and Type 2 fabrics, respectively. Each represents... 111
Fig. 3.24. (a) Contact experiment with angles. (b) Picking up fabric at attack angles of 5° and (c) 10° from the vertical axis. 112
Fig. 3.25. Proposed soft gripper (a) before being pressed, with pressing forces of (b) 25 N and (c) 35 N, (d) 50 N. (e) Pressing force... 113
Fig. 3.26. Experimental results of barometric pressure sensor. (a) Measured force as the height of the gripper changes. (b) Measured... 115
Fig. 3.27. Automatic fabric handling using proposed gripper system. (a) Sequence of picking up air-permeable fabric placed on coated non-... 117
Fig. 4.1. Soft robotic arm made of manipulator modules composed of bellow-type soft actuators. (a) Single- and (b) double-module... 122
Fig. 4.2. Force measurement results. Maximum pulling forces are 1.4 N for string strain sensor and 3.5 N for wire encoder (CWPS1000V1,... 123
Fig. 4.3. (a) Exploded view of manipulator module. (b) Design of the bellow soft actuator and configuration of the major components: i.... 128
Fig. 4.4. Fabrication process. (a) 3D printing of the bellow structure. (b) Plasma treatment. (c) Dipping in HDFS for amplifying the... 131
Fig. 4.5. Fabrication process of soft strain sensor. (a) Direct printing of a liquid-metal trace on a silicone substrate. (b) Encapsulation of the... 132
Fig. 4.6. Two actuation modes of the manipulator module: (a) Linear translation and (b) bending in sequence. 135
Fig. 4.7. Constant curvature model. When the bellows are given with different pressures, the manipulator module bends and forms a... 138
Fig. 4.8. Simulation conditions and the results. (a) Linear translation: (Left) internally pressurized manipulator, and (Right) externally loaded... 141
Fig. 4.9. Experimental result of soft string sensor. (a) Normalized sensor signal during a cycle of stretching and releasing. The blue... 146
Fig. 4.10. String sensor initialization structure. Calibration is performed using a tensile tester externally before connecting to the... 147
Fig. 4.11. Cost function calculation result. The manipulator module repeatedly moves between p₁ (0, 0, 162) and p₂ (0, 30, 140). (a)... 148
Fig. 4.12. Vision-based object detection and central point localization of the top of the manipulator module. (a) Schematic of object detection... 151
Fig. 4.13. Object detection result with different configuration. (a) Photo of the manipulator module is bent to the left with the gripper... 152
Fig. 4.14. Pressure responses of the manipulator module. (a) Range of bending angles with different input pressures. The solid line... 154
Fig. 4.15. Experimental result with various pressure inputs. (a) Step responses of the module. (b) Pressure response with different control... 157
Fig. 4.16. Pressure and touch sensing results when (a) p₁, p₂, and p₃=−10 kPa. (b) p₁=−30 kPa, p₂=−20 kPa, and p₃=−40 kPa. (c)... 160
Fig. 4.17. Depth camera measurement accuracy test setup and result. (a) Schematic of local frames and position vectors for the depth... 162
Fig. 4.18. Manipulator module control results. (a) Results of bellow length feedback control using only vision. (b) The corresponding... 163
Fig. 4.19. Load mounted manipulator module control experiment results. (a) Image of gripper mounted manipulator module (left) and... 167
Fig. 4.20. Goal position adjustments in the presence of touch while controlling the load-mounted manipulator module. Even if a touch... 168
Fig. 4.21. Application using the assembled soft robotic arm with embedded gels and mounted gripper. (a) The top graph plots the state... 169
Fig. 4.22. Fabric handling soft manipulation system. (a) Soft robotic arm equipped with the soft grippers. (b) Soft pressure sensor reading... 171