표제지
목차
국문초록 9
1. 서론 11
1.1. 연구 배경 11
1.2. 연구 동기 및 목적 18
1.3. 논문 구성 20
2. 소실점 검출 및 자율주행 제어 21
2.1. 소실점 검출 21
2.2. 소실점을 이용한 자율주행 차량 조향제어 35
2.2.1. 일반적 자율주행 차량 조향제어 35
2.2.2. 속도를 고려한 자율주행 차량 조향제어 38
2.2.3. 단일변수를 갖는 자율주행 차량 조향제어 44
2.2.4. 차축에서 자율주행 차량 조향제어 47
2.2.5. 보조 PID 제어기 설계 52
2.3. 소실점을 이용한 자율주행 차량 조향제어기 설계 55
3. 자율주행을 위한 CNN 제어기 설계 58
3.1. CNN 58
3.1.1. CNN 구조 60
3.1.2. 조향제어를 위한 CNN의 설계 67
3.2. CNN을 이용한 자율주행 차량 조향제어기 설계 69
4. 모의실험 환경 및 학습 데이터 74
4.1. 모의실험 환경 74
4.2. 테스트 로드 75
4.3. 학습 데이터 80
4.4. 제안한 CNN의 설계 및 학습 82
5. 모의실험 및 결과 97
5.1. 소실점을 이용한 자율주행 차량 성능평가 98
5.1.1. Test road 1에 대한 성능평가 98
5.1.2. Test road 2에 대한 성능평가 102
5.1.3. Test road 3에 대한 성능평가 107
5.2. CNN을 이용한 자율주행 차량 성능평가 108
5.2.1. Test road 1에 대한 성능평가 109
5.2.2. Test road 2에 대한 성능평가 113
5.2.3. Test road 3에 대한 성능평가 118
5.2.4. Test road 4에 대한 성능평가 122
6. 결론 126
참고문헌 130
ABSTRACT 140
Table 1. Machine learning algorithm 15
Table 2. Section speed 43
Table 3. Steering controllers of an autonomous vehicle using vanishing point 56
Table 4. Steering controllers of an autonomous vehicle using CNN 70
Table 5. Image data for CNN learning 81
Table 6. Summary of the proposed CNN model 84
Table 7. Environment settings for simulation 97
Table 8. Performance evaluation of autonomous vehicles using vanishing point(Test road 1) 102
Table 9. Performance evaluation of autonomous vehicles using vanishing point(Test road 2) 106
Table 10. Performance evaluation of autonomous vehicles using CNN(Test road 1) 113
Table 11. Performance evaluation of autonomous vehicles using CNN(Test road 2) 117
Table 12. Performance evaluation of autonomous vehicles using CNN(Test road 3) 121
Table 13. Performance evaluation of autonomous vehicles using CNN(Test road 4) 125
Fig. 1. Typical lane detection algorithm 22
Fig. 2. Vanishing point and lane detection 23
Fig. 3. Camera position and image region captured by the camera 24
Fig. 4. 2-dimensional image region of camera 25
Fig. 5. Camera image 26
Fig. 6. Gray image 27
Fig. 7. Edge transformation according to two thresholds 28
Fig. 8. Canny edge transformation(threshold_H=200, threshold_L=120) 29
Fig. 9. Hough transform 30
Fig. 10. Hough transform of Image 31
Fig. 11. Detect vanishing points according to road environment(threshold=60) 33
Fig. 12. Detect vanishing points according to road environment(threshold=45) 34
Fig. 13. Detect vanishing points according to road environment(threshold=30) 34
Fig. 14. Steering control of autonomous vehicle with constant speed 37
Fig. 15. Block diagram for steering control of autonomous vehicle steering with constant speed 38
Fig. 16. Look-ahead distance 39
Fig. 17. Steering control of autonomous vehicle with variable speed 40
Fig. 18. Block diagram for steering control of autonomous vehicle considering speed 41
Fig. 19. Test road with variable speed section 43
Fig. 20. The error between the autonomous vehicle and the center... 44
Fig. 21. Steering control of autonomous vehicle considering speed with a single parameter 45
Fig. 22. Block diagram for steering control of autonomous vehicle with variable speed... 46
Fig. 23. Comparison of the performance of an autonomous vehicle... 47
Fig. 24. Example of incorrect steering angle according to vehicle position 48
Fig. 25. Side view of an autonomous vehicle equipped with a camera 49
Fig. 26. Steering control of an autonomous vehicle at front-wheel center 50
Fig. 27. Block diagram for steering control of autonomous vehicle with variable speed on... 51
Fig. 28. Comparison of the performance of autonomous vehicles... 52
Fig. 29. Steering control of autonomous vehicle at front-wheel center with PID controller 53
Fig. 30. Block diagram for steering control of autonomous vehicle at front-wheel center... 54
Fig. 31. Comparison of the performance of autonomous vehicles at... 55
Fig. 32. General CNN structure for classification 60
Fig. 33. General CNN structure for regression 68
Fig. 34. Block diagram for steering control of autonomous vehicle using CNN_Method 1 71
Fig. 35. Block diagram for steering control of autonomous vehicle using CNN_Method 2 72
Fig. 36. Block diagram for steering control of autonomous vehicle using CNN_Method 3 72
Fig. 37. Autonomous vehicle in a virtual environment 75
Fig. 38. Road applied in a virtual environment 76
Fig. 39. Roadmap in a virtual environment 77
Fig. 40. The environment around the road (a)~(b) Test road 1 (c)~(d) Test road 2... 79
Fig. 41. Roads whose edges are designed unclearly from the ground 80
Fig. 42. Measurement system for vanishing point, right and left vanishing lines for... 81
Fig. 43. The proposed CNN structure 82
Fig. 44. Structure of the first convolution layer of the proposed CNN 85
Fig. 45. Filters of the first convolution layer 86
Fig. 46. An example of image 87
Fig. 47. Feature map after the first convolution 88
Fig. 48. Feature map after applying the activation function of the first convolution layer 89
Fig. 49. Feature map after applying the pooling of the first convolution layer 90
Fig. 50. Feature map of the fourth convolution layer 92
Fig. 51. Loss and Accuracy (a) Loss to epoch (b) Accuracy to epoch 94
Fig. 52. CNN learning error (a) Error of left vanishing line (b)... 96
Fig. 53. Tracking performance of autonomous vehicles using vanishing point(Test road 1) 100
Fig. 54. Tracking error of autonomous vehicles using vanishing point(Test road 1) 102
Fig. 55. Tracking performance of autonomous vehicles using vanishing point(Test road 2) 104
Fig. 56. Tracking error of autonomous vehicles using vanishing point(Test road 2) 106
Fig. 57. Tracking performance of autonomous vehicles using vanishing point(Test road 3) 107
Fig. 58. Tracking error of autonomous vehicles using vanishing point(Test road 3) 108
Fig. 59. Tracking performance of autonomous vehicles using CNN(Test road 1) 111
Fig. 60. Tracking error of autonomous vehicles using CNN(Test road 1) 113
Fig. 61. Tracking performance of autonomous vehicles using CNN(Test road 2) 115
Fig. 62. Tracking error of autonomous vehicles using CNN(Test road 2) 117
Fig. 63. Tracking performance of autonomous vehicles using CNN(Test road 3) 119
Fig. 64. Tracking error of autonomous vehicles using CNN(Test road 3) 121
Fig. 65. Tracking performance of autonomous vehicles using CNN(Test road 4) 123
Fig. 66. Tracking error of autonomous vehicles using CNN(Test road 4) 125