同步定位與建圖是一項在機器人領域中相當重要的技術，該技術能夠使機器人在特定區域中自主移動。該技術依照感測器不同，可分為視覺或是激光系統。無人機因其體積較小與載重能力較差，經常採取視覺的方式。視覺同步定位與建圖經常面臨定位丟失、誤差累積及計算能力需求較高等問題。在構建地圖的過程中經常受到光線變化、物體遮擋和感知混淆等不確定因素的影響，導致定位失敗。閉迴路偵測是該技術中一個重要的模塊，用於減少地圖構建過程中累積的軌跡誤差，並進行地點偵測。然而該模塊在視覺系統中，經常受到光照、環境或是視角變化的影響，導致機器人無法有效偵測。常見的閉迴路偵測是基於Bag of Words，本文提出一種基於ORB特徵和深度學習的閉迴路偵測方法。主要的目的是提高機器人在面對不同不確定性因素時的地點辨識能力，使用深度學習來偵測環境中之物體，已進行閉迴路偵測。本文通過指標(ATE，RPE，P-R曲線以及F1 score)以及比較不同的同步定位與建圖系統來進行評估。根據實驗結果，而所提出的方式與其他算法相比擁有最低的ATE以及RPE，並且提高閉迴路模塊面對不確定因素時的強健性，同時產生更精確的地圖。

Simultaneous localization and mapping is used to help robots locate itself and navigate in an unknown environment, in the meantime, constructing the map of environment. Visual simultaneous localization and mapping often faced problems such as localization lost, accumulated error and high demand of computation capability. While constructing the map, it often suffers from uncertainties such as illumination changes, object occlusion, and perceptual aliasing lead to localization failed. Loop closure detection is a key module of visual simultaneous localization and mapping. It’s used to reduce the cumulative error while constructing the map and detect the place. Thus, it is a key part determining the accuracy and stability of system. Traditionally, most of the loop closure method were based on the bag of words, converts image into a descriptor and compares it with the descriptor database to determine loop closure. But bag of words only compares the descriptors are matched or not. Therefore, this paper proposed a novel loop closure detection method based on oriented fast and rotated brief feature and deep learning. The purpose is to enhance the capability of robot to recognize the current place against different uncertainties. Used deep learning to detect objects in the environment for loop closure detecting. The deep learning module used you only look once to reduce the time cost. As the result shows, the proposed frame work shows the lowest absolute trajectory error and relative pose error compared to other algorithms. The result of the study shows that the proposed method can raise the robustness of loop closure module facing the uncertainties, in the mean while generate to more precisely environment map.

[1] N. A. Bayanbay, I. K. Beisembetov, K. A. Ozhikenov, O. E. Bezborodova, O. N. Bodin, and V. G. Polosin, “The Use of Unmanned Aerial Vehicle for Emergency Medical Assistance,” in Proc. 20th International Conference of Young Specialists on Micro/Nanotechnologies and Electron Devices (EDM), Erlagol, Russia, 2019, pp. 597-600.
[2] P. Karthigeyan, M. S. Raja, S. Prabu, and R. Gnanaselvam, “Flying robot - A drone for urban warfare,” in Proc. International Conference on Pervasive Computing (ICPC), Pune, India, 2015, pp. 1-4.
[3] S. Çaşka and A. Gayretlı, “An algorithm for collaborative patrolling systems with unmanned air vehicles and unmanned ground vehicles,” in Proc. 7th International Conference on Recent Advances in Space Technologies (RAST), Istanbul, Turkey, 2015, pp. 659-663.
[4] A. Valsan, B. P, H. V. D. G, R. S. Unnikrishnan, P. K. Reddy, and A. V, “Unmanned Aerial Vehicle for Search and Rescue Mission,” in Proc. 4th International Conference on Trends in Electronics and Informatics (ICOEI), Tirunelveli, India, 2020, pp. 684-687.
[5] R. O. E. Mohamed, S. M. A. Mohamed, and H. A. E. Abashar, “Unmanned Aerial Vehicle Intelligent Ambulance Based System,” in Proc. International Conference on Computer, Control, Electrical, and Electronics Engineering (ICCCEEE), Khartoum, Sudan, 2021, pp. 1-5.
[6] Y. A. Gunchenko, S. A. Shvorov, V. I. Zagrebnyuk, V. U. Kumysh, and E. S. Lenkov, “Using UAV for unmanned agricultural harvesting equipment route planning and harvest volume measuring,” in Proc. IEEE 4th International Conference Actual Problems of Unmanned Aerial Vehicles Developments (APUAVD), Kiev, Ukraine, 2017, pp. 262-265.
58
[7] G. D. Georgiev, G. Hristov, P. Zahariev, and D. Kinaneva, “Forest Monitoring System for Early Fire Detection Based on Convolutional Neural Network and UAV imagery,” in Proc. 28th National Conference with International Participation (TELECOM), Sofia, Bulgaria, 2020, pp. 57-60.
[8] T. R. Gayathri, R. P. Aneesh, and G. R. Nayar,“Feature based simultaneous localisation and mapping,” IEEE International Conference on Circuits and Systems (ICCS), Thiruvananthapuram, India, 2017, pp. 419-422.
[9] D. Wang, K. Tan, and H. Li, “Research on feature extraction method based on Simultaneous localization and mapping,” in Proc. 37th Chinese Control Conference (CCC), Wuhan, China, 2018, pp. 3720-3724.
[10] R. Mur-Artal, J. M. M. Montiel, and J. D. Tardós, “ORB-SLAM: A Versatile and Accurate Monocular SLAM System,” IEEE Transactions on Robotics. vol. 31, no. 5, pp. 1147-1163, 2015.
[11] A. Li, X. Ruan, J. Huang, X. Zhu, and F. Wang, “Review of vision-based Simultaneous Localization and Mapping,” in Proc. IEEE 3rd Information Technology, Networking, Electronic and Automation Control Conference (ITNEC), Chengdu, China, 2019, pp. 117-123.
[12] Z. Huishen, X. Ling, Y. Huan, and W. Liujun, “An improved bag of words method for appearance based visual loop closure detection,” in Proc. Chinese Control and Decision Conference (CCDC), Shenyang, China, 2018, pp. 5682-5687.
[13] J. Lai, Z. Liu, and J. Lin, “Loop Closure Detection for Visual SLAM Systems Using Various CNN algorithms Contrasts,” in Proc. Chinese Automation Congress (CAC), Hangzhou, China, 2019, pp. 1663-1668.
[14] S. Lee, H. Jo, H. M. Cho, and E. Kim, “Robust Visual Loop Closure Detection with Repetitive Features,” in Proc. 15th International Conference on Ubiquitous Robots (UR), Honolulu, HI, USA, 2018, pp. 891-895.
59
[15] S. Lee, H. Jo, H. M. Cho, and E. Kim, “Visual Loop Closure Detection over Illumination Change,” in Proc. 16th International Conference on Ubiquitous Robots (UR), Jeju, South of Korea, 2019, pp. 77-80.
[16] P. Yu, X. Ruan, and X. Zhu, “The loop closure Detection Algorithm Based on Bag of Semantic Word For Robot Navigation,” in Proc. IEEE International Conference on Information Technology,Big Data and Artificial Intelligence (ICIBA), Chongqing, China, 2020, pp. 54-58.
[17] K. Zhang and W. Zhang, “Loop closure detection based on generative adversarial networks for simultaneous localization and mapping systems,” in Proc. Chinese Automation Congress (CAC), Jinan, China, 2017, pp. 7916-7919.
[18] K. Quan, B. Xiao, and Y. Wei, “Intelligent Descriptor of Loop Closure Detection for Visual SLAM Systems,” in Proc. Chinese Control and Decision Conference (CCDC), Nanchang, China, 2019, pp. 993-997.
[19] H. Li, C. Tian, L. Wang, and H. Lv, “A loop closure detection method based on semantic segmentation and convolutional neural network,” in Proc. International Conference on Artificial Intelligence and Electromechanical Automation (AIEA), Guangzhou, China, 2021, pp. 269-272.
[20] C. Campos, R. Elvira, J. J. G. Rodríguez, J. M. M. Montiel, and J. D. Tardós, “ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual–Inertial, and Multimap SLAM,” IEEE Transactions on Robotics. vol. 37, no. 6, pp. 1874-1890, 2021.
[21] E. Rublee, V. Rabaud, K. Konolige, and G. Bradski, “ORB: An efficient alternative to SIFT or SURF,” in Proc. 2011 International Conference on Computer Vision, 2011, pp. 2564-2571.
[22] D. Galvez-López and J. D. Tardos, “Bags of Binary Words for Fast Place Recognition in Image Sequences,” IEEE Transactions on Robotics. vol. 28, no. 5, pp. 1188-1197, 2012.
60
[23] T. Y. Lin, M. Marie, S. Belongie, J. Hays, P. Perona, D. Ramanan and C. L. Zitnick, “Microsoft coco: Common objects in context,” in Proc. European conference on computer vision, Springer, Cham, 2014, pp. 740-755.
[24] Y. Li, Q. Wang, and R. Liu, “Research on YOLOv3 pedestrian detection algorithm based on channel attention mechanism,” in Proc. 2021 IEEE International Conference on Computer Science, Electronic Information Engineering and Intelligent Control Technology (CEI), 2021, pp. 229-232.
[25] B. Bescos, J. M. Fácil, J. Civera, and J. Neira, “DynaSLAM: Tracking, Mapping, and Inpainting in Dynamic Scenes,” IEEE Robotics and Automation Letters. vol. 3, no. 4, pp. 4076-4083, 2018.
[26] D. Frost, V. Prisacariu, and D. Murray, “Recovering Stable Scale in Monocular SLAM Using Object-Supplemented Bundle Adjustment,” IEEE Transactions on Robotics. vol. 34, no. 3, pp. 736-747, 2018.
[27] C. Dang-Nguyen and T. Do-Hong, “Robust Line Hausdorff Distance for Face Recognition,” in Proc. 2019 International Symposium on Electrical and Electronics Engineering (ISEE), 2019, pp. 103-107.
[28] L. Wang, X. Mu, C. Ma, and J. Zhang, “Hausdorff IoU and Context Maximum Selection NMS: Improving Object Detection in Remote Sensing Images With a Novel Metric and Postprocessing Module,” IEEE Geoscience and Remote Sensing Letters. vol. 19, no. pp. 1-5, 2022.
[29] M. Zhu and L. Huang, “Fast and Robust Visual Loop Closure Detection with Convolutional Neural Network,” in Proc. 2021 IEEE 3rd International Conference on Frontiers Technology of Information and Computer (ICFTIC), 2021, pp. 595-598.
[30] S. Arshad and G. W. Kim, “Semantics Aware Loop Closure Detection in Visual SLAM,” in Proc. 21st International Conference on Control, Automation and Systems (ICCAS), Jeju, Republic of Korea, 2021, pp. 21-24.
61
[31] Z. Qian, J. Fu, and J. Xiao, “Towards Accurate Loop Closure Detection in Semantic SLAM With 3D Semantic Covisibility Graphs,” IEEE Robotics and Automation Letters. vol. 7, no. 2, pp. 2455-2462, 2022.