傅立成臺灣大學：電機工程學研究所吳俊逸Wu, Chun-YiChun-YiWu2007-11-262018-07-062007-11-262018-07-062007http://ntur.lib.ntu.edu.tw//handle/246246/53091在大尺度環境下機器人同步建構地圖與自我定位(Simultaneously Localization and Mapping, SLAM)的應用中，常會遇到一個問題，即過多的地圖地標無可避免地會讓共同估測機器人位置與地圖地標的濾波器計算負擔太大。這主要是由於兩個原因所造成：一方面是在地圖地標的選擇機制上不夠扎實，導致在環境觀測過程中不必要的定位地標太多；而另ㄧ方面，則是濾波器的本身數學特性，導致計算負擔的增加。在本論文中，我們提出了一個結合加速強健特徵擷取(SURF Extraction)以及逆深度特徵初始化(Inverse Depth Initialization)的影像前端系統，來有效的選出強健的靜止地圖地標，用以提供定位及地圖資訊，並且在已知地圖再次觀測到的前提之下，有效達到大範圍的不確定縮減。此外，在後端濾波器的選擇上，我們將稀疏線性化資訊濾波器演算法延伸到影像感測器的應用。稀疏線性化資訊濾波器已被證實，在使用雷射實現SLAM時，可以有效的維持計算效能。最後，透過實驗以及模擬，我們證實了此系統的效能及可靠性。In the application of root Simultaneous Localization and Mapping (SLAM) in a large scale environment, it remains a challenge to resolve the obstacle of the inevitable computational burden on the filtering scheme imposed by the excessive number of landmarks. This obstacle maily attributes to two facts: one is that the selection scheme is not sufficiently stringent, thus resulting in the inclusion of valueless localization landmarks during the environment observation process; the other is the mathematical characteristic of the filter, i.e. the computational complexity is proportional to the number of landmarks. In this thesis, we propose a visual front-end system integrating the speed-up robust feature extraction (SURF Extraction) and Inverse Depth Initialization to efficiently and effectively select robust static landmark for the information of localization and mapping and significantly reduce the uncertainty of the large exploration environment under the presumption of re-observation of the map. Furthermore, we extend the sparse linearization information filtering algorithm to the application of visual sensor. In the SLAM of laser, it has been proved the adoption of sparse linearization information filter effectively improve the computational efficiency. The performance and reliability is validated by the simulation and experiments.Chapter 1 Introduction 1 1.1 Motivation 1 1.2 Literature Review 2 1.3 Contribution 3 1.4 Thesis Organization 4 Chapter 2 Preliminaries 5 2.1 Fundamental of SLAM 5 2.1.1 Bayesian Method 5 2.1.2 Feature-based approach in SLAM 7 2.1.3 Extended Kalman Filter (EKF) in SLAM 9 2.1.4 Extended Information Filter (EIF) in SLAM 17 2.2 SLAM Using Vision 19 2.2.1 Top-down Approaches 20 2.2.2 Bottom-up Approaches 23 2.3 Feature Extraction in Computer Vision 23 2.3.1 Feature Detection 24 2.3.2 Feature Description 26 2.4 Landmark Initialization 28 2.4.1 Delayed Approaches 29 2.4.2 Non-delayed approaches 30 Chapter 3 Visual Front-end System 35 3.1 System Diagram of Visual Front-end System 38 3.2 Speeded Up Robust Feature (SURF) 45 3.2.1 Fast Hessian Detector 45 3.2.2 SURF Descriptor 46 3.3 Algorithm for deleting features on moving objects 48 3.4 Database Management for Place Recognition 49 Chapter 4 System Modeling and Filter Design 51 4.1 System Diagram 51 4.2 Motion Model 52 4.2.1 2D Robot Motion Model 52 4.2.2 2.5D Camera Model on Robot Platform 55 4.3 Measurement Model 56 4.4 Sparse Extended Information Filter in SLAM 57 Chapter 5 Simulation and Experiment Results 65 Introduction of Experimental Equipment and Environment 65 5.1 65 5.1.1 Hardware of the Experimental System (Tour Guide Robot) 65 5.1.2 Hardware of the Experimental System (Pioneer) 66 5.1.3 Experimental Environment 67 Performance Evaluation in Visual Front-end System 68 5.2 68 5.2.1 The performance of the inverse depth landmark initialization 68 5.2.2 The performance of the visual front end system 70 Total Results of the Integrated System 72 5.3 72 Chapter 6 Conclusion and Future Work 76 6.1 Conclusion 76 6.2 Future work 763402052 bytesapplication/pdfen-US機器人同步建構地圖與自我定位影像特徵擷取與初始化影像前端系統稀疏線性化資訊濾波器robot SLAMfeature extraction and initializationvisual front-end systemsparse linearization information filterthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/53091/1/ntu-96-R94921014-1.pdf