指導教授：羅仁權臺灣大學：電機工程學研究所吳謝浩Wu, XiehaoXiehaoWu2014-11-282018-07-062014-11-282018-07-062014http://ntur.lib.ntu.edu.tw//handle/246246/262943本論文主旨在於發展在室內環境可以同時完成定位地圖構建 (SLAM) 與移動物體追蹤功能的行動機器人。SLAM可以幫助機器人定位與構建周圍環境、移動物體追蹤功能則把外界環境區分為靜止部份和移動部份。根據靜態地圖能夠偵測移動物體，同時在追蹤移動物體時又能夠幫助區分物體是否為靜態物。在機器人以及移動物體軌跡優化的框架下，本文把基於圖優化的SLAM延伸到了基於圖優化的SLAM與移動物體追蹤，從而同時對機器人軌跡以及移動物體軌跡進行優化。基於移動物體的量測，對移動物體未來行為以及過去行為進行推測，使得對於移動物體不同時刻的觀測可以互相幫助，從而得到更好的移動物體軌跡估測。對於機器人在複雜室內環境實現SLAM與移動物體追蹤會碰到的難題，例如移動物體的形狀和特點各異，很難利用先驗知識去建模，並且在複雜室內環境進行數據關聯也十分困難。本文提出的基於長數據序列的移動物體偵測在不需要對移動物體有先驗知識的情況下，完成移動物體偵測。在此方法下即使是十分細微的移動也會被偵測出來。另外，多感測器融合方法提高了數據關聯的精准度。實驗結果證明了本文的多感測器融合基於圖優化SLAM與移動物體追蹤方法可以在複雜室內環境中實現，把移動物體整合到基於圖優化SLAM中對比不整合移動物體的圖優化SLAM，降低了機器人姿態估測的不確定性。The objective of this thesis is to develop simultaneous localization and mapping (SLAM) with capability of tracking moving object in indoor environments. SLAM can help build environment map, while detection and tracking of moving object separate the environment into static and dynamic parts. The map can help detect the moving object, on the other hand, the moving object tracking can help separate the stationary and moving objects, thus we can separate them in the map. By augmenting the moving objects state and related constraints into the robot and objects graph, the general graph-based framework for SLAM issues can be extended to jointly optimize the SLAM and moving object tracking result. By incorporating the moving object prediction and moving object Retro-BestGuess, the later measurement of moving object can help the estimation of the previous state and vice versa. Consequently, the trajectory of robot together with the trajectories of moving objects is optimized. Furthermore, the SLAM with moving object tracking issues in the cluttered indoor environment are analyzed, the moving object may have different size and characteristics difficult to modelling, and the data association is difficult. The multi-frame moving object detection is applied to detect the moving object without the need of prior knowledge, by which even the slightly movement can be detected. The multi-sensor fusion methodologies can help increase the data association accuracy. The experimental results shown that our algorithm is feasible in cluttered indoor environment, graph-based SLAM incorporating moving objects can decrease the pose estimation uncertainty compare to the one not incorporating them.口試委員會審定書 # 誌謝 i 中文摘要 ii ABSTRACT iii CONTENTS v LIST OF FIGURES ix Chapter 1 Introduction 1 1.1 Scene Understanding for Indoor Service Robot 2 1.1.1 Localization 3 1.1.2 Simultaneous Localization and Mapping 3 1.1.3 Moving Objects Tracking 5 1.1.4 SLAM vs. Moving Object Tracking 6 1.2 Related Research 7 1.3 Contributions of This Thesis 9 1.4 Experimental Setup 11 1.5 Organization 13 Chapter 2 Probabilistic Foundations 15 2.1 Uncertain Spatial Relationships 16 2.1.1 Compounding 16 2.1.2 The Inverse Relationship 18 2.1.3 The Tail-to-Tail Relationship 19 2.2 Graph-based Simultaneous Localization and Mapping 20 2.2.1 SLAM as a Belief Net 20 2.2.2 SLAM as a Factor Graph 21 2.2.3 SLAM as a Markov Random Field 23 2.2.4 Filtering-based SLAM VS Smoothing-based SLAM 23 2.2.5 Graph-based SLAM as a Least Squares Problem 26 2.2.6 Perception Modelling and Data Association 29 2.3 Graph-based Moving Object Tracking 29 2.3.1 Formulation of Moving Object Tracking 29 2.3.2 Graph-based Smoothing Framework for Moving Object Tracking 31 2.3.3 Motion Modelling 33 2.3.4 Perception Modelling and Data Association 33 2.4 Graph-based SLAM with Moving Object Tracking 34 2.4.1 Formulation of SLAM with Moving Object Tracking 34 2.4.2 Graph-based SLAM with moving object tracking VS Filtering-based SLAM with moving object tracking 36 2.5 Summary 37 Chapter 3 Perception Modelling 39 3.1 Object Representation in the Cluttered Indoor Environment 39 3.1.1 Scan Segmentation 40 3.1.2 Perception Sensor Modelling 40 3.1.3 Inferred velocity from uncertain position 41 3.2 Range Scan Matching 42 3.2.1 The Iterated Closest Point Algorithm 42 3.3 Moving Object Representation for Tracking 44 3.4 Map Representation of Graph-based SLAM with moving object tracking 46 3.5 Summary 47 Chapter 4 Motion Modelling 48 4.1 Robot Motion Modelling 48 4.2 Moving Object Motion Modelling 49 4.2.1 The Constant Velocity Model 49 4.2.2 Motion Retro-BestGuess 50 4.3 Summary 51 Chapter 5 Data Association 52 5.1 Multi-sensory Fusion Data Association of Moving Object 52 5.2 Multi-sensory Fusion Loop Closure Detection 55 5.3 Summary 56 Chapter 6 Implementation 57 6.1 Process of Implementation 57 6.2 Multi-Frame Moving Object Detection 59 6.2.1 Multiple Interval Frame Occupancy Grid-based Detection 60 6.2.2 Near Frame Object Association 61 6.3 Experimental Results 62 6.3.1 Moving Object Detection 63 6.3.2 Mapping in dynamic environment 64 6.3.3 Graph-based SLAM with moving object tracking 64 6.3.4 Quantitative analysis 69 6.4 2D Assumption Failure 70 6.5 Summary 72 Chapter 7 Conclusion 73 Chapter 8 Future Extensions 76 REFERENCE 78 VITA 825035345 bytesapplication/pdf論文公開時間：2015/08/08論文使用權限：同意有償授權(權利金給回饋學校)同時定位地圖構建移動物體追蹤軌跡優化多感測器融合複雜室內環境thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/262943/1/ntu-103-R01921082-1.pdf