導航控制是自主式移動機器人的核心功能之一，其可讓機器人在工作環境中完成移動控制及迴避障礙物的任務。現有的導航控制技術大多是基於已知的環境地圖來進行，若在未知環境中或沒有當前的環境地圖情況下，則機器人需要先進行地圖建置程序後，才能開始執行導航控制任務。為了克服此限制，本論文提出一種基於深度模仿學習的雙輪移動機器人之無地圖式光達導航控制系統，其直接使用光達感測器資訊與目標點座標資訊進行數據驅動控制。透過所提出的深層卷積網路模型，即可輸出移動控制命令，且不需要環境地圖資訊及調整導航演算法的參數，即可達成在動態或未知環境中移動機器人導航控制。在訓練數據集的收集上，我們透過人工操控方式，操控雙輪移動機器人進行避障移動，並將光達感測器資訊、目標點相對座標資訊及移動控制命令記錄下來，且透過資料擴增來增加數據集的數據數量。在網路模型設計中，所提出的CNN模型包括一個光達訊號卷積模塊與一個移動預測模塊，用來提取光達資訊特徵及機器人的移動行為預測。在模型訓練中，透過我們人工給予的專家策略，將輸入的光達感測器資訊與目標點座標資訊經過端到端模仿學習映射出移動控制命令。實驗結果顯示，所提出的雙輪移動機器人之無地圖式光達導航控制系統可以在現有的環境中安全地導航，也能在未知環境中或沒有當前的環境地圖情況中達到80%的成功率導航到目標點座標，其導航效果與專家策略相近證明提出的系統可以克服此限制。

Navigation control is one of the core functions of autonomous mobile robots, which allows the robot to complete the task of motion control and avoiding obstacles in the working environment. Most of the existing navigation control technologies are based on known environment maps. If the robot is in an unknown environment or there is no available environment map, the robot needs to build a map before it can start to perform navigation control tasks. In order to overcome this limitation, this thesis proposes a mapless LiDAR navigation control of two-wheeled mobile robots based on deep imitation learning, which directly uses LiDAR sensor information and target point coordinate information for data-driven navigation control. The proposed deep convolutional neural network model can output motion control commands without the requirement of environment map information and the adjustment of the parameters of the navigation algorithm to achieve navigation control of the mobile robot in a dynamic or unknown environment. In the collection of the training dataset, we manipulated the two-wheeled mobile robot to avoid obstacles through manual control and recorded the information of the LiDAR sensor, the relative coordinate information of the target point, and the motion control commands. Next, we applied a data augmentation method on the recorded samples to increase the number of training samples in the dataset. In the network model design, the proposed CNN model includes a LiDAR signal convolution neural network module and a movement prediction module, which are used to extract LiDAR information features and predict the motion behavior of the robot, respectively. In the model training phase, the proposed CNN model learns how to map the input LiDAR sensor information and target point coordinate information to the motion control command through end-to-end imitation learning. Experimental results show that the proposed mapless LiDAR navigation control of two-wheeled mobile robots system can safely navigate in the known environment, and can also navigate to the target point with a success rate of 80% in an unknown environment without the environment map. These experimental results validate that the proposed mapless LiDAR navigation control system can overcome the limitation of navigation control in an unknown environment without the environment map.

中文摘要	I
英文摘要	III
目錄	V
圖目錄	VIII
表目錄	X
第一章 序論	1
1.1  研究背景	1
1.2  研究動機與目的	4
1.3  系統架構	5
1.4  論文貢獻	8
1.5  論文架構	9
第二章 相關研究	10
2.1  數據驅動	10
2.1.1視覺影像特徵方法	10
2.1.2 光達點雲特徵方法	11
2.2  卷積類神經網路	12
2.3  無地圖光達導航控制方法	13
2.3.1 基於點雲處理	14
2.3.2 基於強化學習的移動行為分類	15
2.3.3 基於強化學習的移動行為分析	16
2.3.4基於模仿學習的移動行為分析	17
2.4  文獻總結	18
第三章 光達訊號卷積類神經網路模型與移動預測類神經網路模型	19
3.1  光達訊號卷積類神經網路模型	19
3.1.1 卷積模塊	19
3.1.2 訊號預處理	21
3.2  移動預測類神經網路模型	22
3.2.1 所提出之移動預測類神經網路模型	22
3.2.2 網路輸入資訊預處理	24
3.3  模型訓練	25
第四章 訓練數據集	27
4.1  訓練數據集的蒐集	27
4.2  訓練數據集的處理	28
4.3  訓練數據集的資料擴增	31
第五章 實驗結果與分析	34
5.1  軟硬體介紹	34
5.2  實驗方法與結果	35
5.2.1 已知環境實驗結果	36
5.2.2 未知環境實驗結果	42
5.3實驗小結：	62
第六章 結論與未來展望	63
參考文獻	65
 
圖目錄
圖1.1導航方法比較：(a)、現有導航控制方法架構圖，(b)、無地圖式導航控制方法架構圖	3
圖1.2、高速圖形運算處理器Nvidia-RTX 3090	4
圖1.3、本論文所提之基於深度模仿學習的雙輪移動機器人光達導航控制系統架構圖	7
圖1.4、本論文所提之導航控制類神經網路模型	8
圖2.1、無地圖光達導航控制方法種類的分支圖	13
圖3.1、本論文所使用之光達訊號卷積類神經網路模型	20
圖3.2、本論文之訊號預處理流程圖	22
圖3.3、本論文移動預測類神經網路模型架構圖	23
圖3.4、本論文訓練的過程	25
圖4.1、本論文的實驗場域	28
圖4.2、本論文所使用之雙輪移動機器人(P3-DX)控制動作	30
圖4.3、所提出之第一種資料擴增方法，其應用仿射變換方法來增加數據多樣性，而移動控制命令皆與原始圖一樣	32
圖4.4、所提出之第二種資料擴增方法，其根據目標點相對座標命令的數據進行資料合併動作，以達到資料擴增的目的	33
圖5.1、左側為P3-DX、右側為Velodyne-16	35
圖5.2、在已知環境中，目標點座標位於感測範圍內導航成功的實驗過程	38
圖5.3、在已知環境中，目標點座標位於感測範圍外導航成功的實驗過程	41
圖5.4、在未知環境中，目標點座標位於感測範圍內導航成功的實驗過程	44
圖5.5、在未知環境中，目標點座標位於感測範圍內導航失敗的實驗過程	45
圖5.6、在未知環境中，目標點座標位於感測範圍外導航成功的實驗過程	48
圖5.7、在未知環境中，目標點座標位於感測範圍外導航失敗的實驗過程	49
圖5.8、在差異更大的未知環境中，目標點座標位於感測範圍內導航成功的實驗過程	52
圖5.9、在差異更大的未知環境中，目標點座標位於感測範圍外導航成功的實驗過程	54
圖5.10、將不同維度的目標點相對座標資料輸入進訓練模型計算其絕對平均誤差結果圖……………………………………………………..56
表目錄
表4.1、本論文所使用之雙輪移動機器人控制動作命令	30
表4.2、訓練數據集資料數量前後變化情形表	33
表5.1、本論文使用到的硬體規格表	34
表5.2、本論文使用到的軟體規格表	35
表5.3、在已知環境中，目標點座標位於感測範圍內的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	37
表5.4、在已知環境中，目標點座標位於感測範圍內的實驗環境裡，雙輪移動機器人成功導航的輸出狀態	37
表5.5、在已知環境中，目標點座標位於感測範圍外的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	39
表5.6、在已知環境中，目標點座標位於感測範圍外的實驗環境裡，雙輪移動機器人成功導航的輸出狀態	40
表5.7、在未知環境中，目標點座標位於感測範圍內的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	42
表5.8、在未知環境中，目標點座標位於感測範圍內的實驗環境裡，雙輪移動機器人成功導航的輸出狀態	43
表5.9在未知環境中，目標點座標位於感測範圍內的實驗環境裡，移動機器人導航失敗的輸出狀態	44
表5.10、在未知環境中，目標點座標位於感測範圍外的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	46
表5.11、在未知環境中，目標點座標位於感測範圍外的實驗環境裡，移動機器人導航成功的輸出狀態	47
表5.12、在未知環境中，目標點座標位於感測範圍外的實驗環境裡，移動機器人導航失敗的輸出狀態	48
表5.13、在差異更大的未知環境中，目標點座標位於感測範圍內的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	50
表5.14、在差異更大的未知環境中，目標點座標位於感測範圍內的實驗環境裡，雙輪移動機器人成功導航的輸出狀態	51
表5.15、在差異更大的未知環境中，目標點座標位於感測範圍外的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	53
表5.16、在差異更大的未知環境中，目標點座標位於感測範圍外的實驗環境裡，移動機器人導航成功的輸出狀態	53
表5.17、將不同維度的目標點相對座標資料輸入進訓練模型計算其絕對平均誤差結果	55
表5.18、在已知環境中，目標點座標位於感測範圍內的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	57
表5.19、在已知環境中，目標點座標位於感測範圍外的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	57
表5.20、在未知環境中，目標點座標位於感測範圍內的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	58
表5.21、在未知環境中，目標點座標位於感測範圍外的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	59
表5.22、在差異更大的未知環境中，目標點座標位於感測範圍內的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	60
表5.23、在差異更大的未知環境中，目標點座標位於感測範圍外的實驗環境裡，移動機器人進行五次實驗的導航輸出狀態	61

[1]	W. Yuan, Z. Li, and C. Y. Su, “RGB-D Sensor-based Visual SLAM for Localization and Navigation of Indoor Mobile Robot,” in 2016 International Conference on Advanced Robotics and Mechatronics, Macau, China, Oct. 2016, doi: 10.1109/ICARM.2016.7606899
[2]	R. Liu, J. Shen, C. Chen, and J. Yang, “SLAM for Robotic Navigation by Fusing RGB-D and Inertial Data in Recurrent and Convolutional Neural Networks,” in 2019 IEEE 5th International Conference on Mechatronics System and Robots, Singapore, Sep. 2019, doi: 10.1109/ICMSR.2019.8835472
[3]	X. Liu, B. Guo, and C. Meng, “A Method of Simultaneous Location and Mapping Based on RGB-D Cameras,” in 2016 14th International Conference on Control, Automation, Robotics and Vision, Phuket, Thailand, Nov. 2016, doi: 10.1109/ICARCV.2016.7838786
[4]	Y. Deng, Y. Shan, Z. Gong, and L. Chen, “Large-Scale Navigation Method for Autonomous Mobile Robot Based on Fusion of GPS and Lidar SLAM,” in 2018 Chinese Automation Congress, Xi'an, China, Dec. 2018, doi: 10.1109/CAC.2018.8623646
[5]	X. Hu, M. Wang, C. Qian, C. Huang, and Y. Xia; M. Song, “Lidar-based SLAM and Autonomous Navigation for Forestry Quadrotors,” in 2018 IEEE CSAA Guidance, Navigation and Control Conference, Xiamen, China, Aug. 2018, doi: 10.1109/GNCC42960.2018.9018923
[6]	J. Li; J. Zhao; Y. Kang; X. He; C. Ye, and L. Sun, “Navigation Control Design of a Mobile Robot by Integrating Obstacle Avoidance and LiDAR SLAM,” in 2019 IEEE Intelligent Vehicles Symposium, Paris, France, June 2019, doi: 10.1109/IVS.2019.8813868
[7]	“中光電公司所開發的自主移動機器人使用的SLAM系統.” Available online: https://www.coretronic-robotics.com/tw/product2
[8]	“waymo.” Available online: https://waymo.com/
[9]	K. Song, Y. H. Chiu, L. R. Kang, S. H. Song, C. A. Yang, and P. C. Lu, “Navigation Control Design of a Mobile Robot by Integrating Obstacle Avoidance and LiDAR SLAM,” in IEEE International Conference on Systems, Man, and Cybernetics, Miyazaki, Japan, Oct. 2018, doi: 10.1109/SMC.2018.00317
[10]	 H. Wang, Y. Zhang, Q. Yuan, and H. Wu, “Intelligent Vehicle Visual Navigation System Design,” in International Conference on Machine Vision and Human-machine Interface, Kaifeng, China, Apr. 2010, doi: 10.1109/MVHI.2010.159
[11]	 T. Feng, and X. T. Zeng, “An Ultrasonic Navigation Automatic Guided Vehicle System Used in CIMS,” in Proceedings of the 4th World Congress on Intelligent Control and Automation, Shanghai, China, June 2002, doi: 10.1109/WCICA.2002.1019993
[12]	 M. Filipenko, and I. Afanasyev, “Comparison of Various SLAM Systems for Mobile Robot in an Indoor Environment,” in 2018 International Conference on Intelligent Systems, Funchal - Madeira, Portugal, Sep. 2018, doi: 10.1109/IS.2018.8710464
[13]	 I. Z. Ibragimov, and I. M. Afanasyev, “Comparison of ROS-based Visual SLAM Methods in Homogeneous Indoor Environment,” in 14th Workshop on Positioning, Navigation and Communications, Bremen, Germany, Oct. 2017, doi: 10.1109/WPNC.2017.8250081
[14]	 P. Abbeel, D. Dolgov, A. Y. Ng, and S. Thrun, “Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation,” in IEEE/RSJ International Conference on Intelligent Robots and Systems, Nice, France, Oct. 2008, doi: 10.1109/IROS.2008.4651222
[15]	 Y. Morales, N. Akai, and H. Murase, “Personal Mobility Vehicle Autonomous Navigation Through Pedestrian Flow: A Data Driven Approach for Parameter Extraction,” in 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems, Madrid, Spain, Oct. 2018, doi: 10.1109/IROS.2018.8593902
[16]	 Q. Wang, L. Chen, B. Tian, W. Tian, L. Li, and D. Cao, “End-to-End Autonomous Driving: An Angle Branched Network Approach,” in IEEE Transactions on Vehicular Technology vol. 68, pp. 11599 - 11610, June 2019, doi: 10.1109/TVT.2019.2921918
[17]	 H. Kanayama, T. Ueda, H. Ito, and K. Yamamoto, “Two-mode Mapless Visual Navigation of Indoor Autonomous Mobile Robot using Deep Convolutional Neural Network,” in IEEE/SICE International Symposium on System Integration, Honolulu, USA, Jan. 2020, doi: 10.1109/SII46433.2020.9025851
[18]	 T. Xue, and H. Yu, “Model-agnostic Metalearning-based Text-driven Visual Navigation Model for Unfamiliar Tasks,” IEEE Access, vol. 8, pp. 166742-166752, Sept. 2020, doi: 10.1109/ACCESS.2020.3023014
[19]	 Y. Zhu, R. Mottaghi, and E. Kolve,” Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learning,” Sept. 2016, arXiv:1609.05143.
[20]	 P. Nikdel, M. Chen, and R. Vaughan, “Recognizing and Tracking High-Level, Human-Meaningful Navigation Features of Occupancy Grid Maps,” in 17th Conference on Computer and Robot Vision, Ottawa, Canada, May 2020, doi: 10.1109/CRV50864.2020.00017
[21]	 Y. Lu, W. Wang, and L. Xue, “A Hybrid CNN-LSTM Architecture for Path Planning of Mobile Robots in Unknow Environments,” in Chinese Control And Decision Conference, Hefei, China, Aug. 2020, doi: 10.1109/CCDC49329.2020.9164775
[22]	 H. Kretzschmar, M. Spies, and C. Sprunk, "Socially Compliant Mobile Robot Navigation Via Inverse Reinforcement Learning", International Journal of Robotics Research, vol. 35, no. 11, pp. 1289-1307, 2016.
[23]	 S. Ioffe, and C. Szegedy, “Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift,” in Proc. of the 32nd Int. Conf. on Int. Conf. on Mach. Learn., vol. 37, pp. 448-456, July 2015.
[24]	 G. E. Hinton, N. Srivastava, A. Krizhevsky, I. Sutskever, and R. R. Salakhutdinov, “Improving Neural Networks by Preventing Co-adaptation of Feature Detectors,” June 2012, arXiv:1207.0580.
[25]	 K. He, X. Zhang, S. Ren, and J. Sun, “Deep Residual Learning for Image Recognition,” in IEEE Conf. on Comput. Vision and Pattern Recognit., Las Vegas, NV, USA, 2016, doi: 10.1109/CVPR.2016.90.
[26]	 K. He, X. Zhang, S. Ren, and J. Sun, “Identity Mappings in Deep Residual Networks,” July 2016, arXiv:1603.05027.
[27]	 T. Ort, L, Paull, and D. Rus, “Autonomous Vehicle Navigation in Rural Environments without Detailed Prior Maps,” in IEEE International Conference on Robotics and Automation, Brisbane, QLD, Australia , Sept. 2018, doi: 10.1109/ICRA.2018.8460519
[28]	 Y. Zhang, J. Wang, X. Wang, and J. M. Dolan, “Road-Segmentation Based Curb Detection Method for Self-driving via a 3D-LiDAR Sensor,” in IEEE Transactions on Intelligent Transportation Systems, vol. 8, pp. 3981-3991, Feb. 2018, doi: 10.1109/TITS.2018.2789462
[29]	 Y. Li, “Deep Reinforcement Learning: An Overview,” Jan. 2017, arXiv:1904.04671.
[30]	 L. Qiang, D. Nanxun, L. Huican, and W. Heng, “A Model-free Mapless Navigation Method for Mobile Robot using Reinforcement Learning,” in Chinese Control And Decision Conference, Shenyang, China, July 2018, doi: 10.1109/CCDC.2018.8407713
[31]	 S. Hossain, O. Doukhi, Y. Jo, and D. J. Lee, “Deep Reinforcement Learning-based ROS-Controlled RC Car for Autonomous Path Exploration in the Unknown Environment,” in 20th International Conference on Control, Automation and Systems, Busan, Korea, Dec. 2020, doi: 10.23919/ICCAS50221.2020.9268370
[32]	 B. Zuo, J. Chen, L. Wang, and Y. Wang, “A Reinforcement Learning Based Robotic Navigation System,” in Proc. IEEE Int. Conf. Syst., Man, Cybern., pp. 3452–3457, 2014
[33]	 W. Khaksar, M. Z. Uddin, and J. Torresen, “Learning from Virtual Experience: Mapless Navigation with Neuro-Fuzzy Intelligence,” in International Conference on Intelligent Systems, Funchal-Madeira, Portugal, May 2019, doi: 10.1109/IS.2018.8710525
[34]	 L. Tai, G. Paolo, and M. Liu, “Virtual-to-real Deep Reinforcement Learning: Continuous Control of Mobile Robots for Mapless Navigation,” in IEEE/RSJ International Conference on Intelligent Robots and Systems, Vancouver, BC, Canada, Dec. 2017, doi: 10.1109/IROS.2017.8202134
[35]	 M. Pfeiffer, U. Schwesinger, H. Sommer ,E. Galceran, and R. Siegwart, "Predicting Actions to Act Predictably: Cooperative Partial Motion Planning with Maximum Entropy Models," in 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 2096-2101, Dec. 2016, doi: 10.1109/IROS.2016.7759329
[36]	 M.Pfeiffer, S. Shukla, M. Turchetta, C. Cadena, A. Krause, and R. Siegwart, “Reinforced Imitation: Sample Efficient Deep Reinforcement Learning for Mapless Navigation by Leveraging Prior Demonstrations,” in IEEE Robotics and Automation Letters vol. 3, pp. 4423 - 4430, Sep. 2018, doi: 10.1109/LRA.2018.2869644
[37]	 Hussein, Ahmed, Gaber, M. Medhat, Elyan, Eyad, Jayne, and Chrisina, “Imitation Learning: A Survey of Learning Methods,” in ACM Computing Surveys, Apr. 2017, doi: 10.1145/3054912
[38]	 M. Pfeiffer, M. Schaeuble, J. Nieto, R. Siegwart, and C. Cadena, “From Perception to Decision: A Data-driven Approach to End-to-End Motion Planning for Autonomous Ground Robots,” IEEE International Conference on Robotics and Automation, Singapore, pp. 1527-1533, July 2017, doi: 10.1109/ICRA.2017.7989182
[39]	 M. Hamandi, M. D’Arcy, and P. Fazli, “DeepMoTion: Learning to Navigate Like Humans,” in 2019 28th IEEE International Conference on Robot and Human Interactive Communication, Singapore, New Delhi, India, Oct. 2019, doi: 10.1109/RO-MAN46459.2019.8956408
[40]	 A. F. Agarap, “Deep Learning using Rectified Linear Units,” Mar. 2018, arXiv:1803.08375.