慣性測量單元其體積小且價格便宜，已成為室內定位不可或缺的一部份。為了提供精度準確且能供室內定位使用的服務，行人航位推算是目前多數人致力於研究的技術，然而，此方法不僅容易受到環境及噪聲影響造成誤差累積，也受限於許多變數，例如使用者身高或是手機攜帶方式，目前仍沒有有效的技術克服此缺陷。
為了解決這些問題，本論文提出一個慣性深度神經網路之架構，基於長短期記憶網路結合自注意力機制，並且使用不確定性加權優化模型，藉由平移與旋轉的相對姿態估計行人軌跡。最重要的是，該方法不需要使用者個人信息，也不受限於設備攜帶方式，便可以直接利用慣性測量單元獲得之六維數據，重建準確且可靠的運動軌跡。並且，它不只能用於週期性運動模式，也能用於非週期性運動軌跡。

The inertial measurement units have become an indispensable part of indoor positioning owning to their characteristics of smaller size and cheaper price. To fulfill the objective of making indoor positioning more accurate, pedestrian dead reckoning recently has become a technology that most people are researching. However, this method is susceptible to cumulative error caused by numerous variables such as environmental impacts, noise effects, even the user's height or the ways to carry cellphone. Thus, currently there is still no efficacious technique to overcome this shortcoming.
In order to ameliorate this situation, this thesis proposes an inertial deep neural network architecture, based on a Long Short-Term Memory combined with a self-attention mechanism, uses an uncertainty-weighting optimization model to estimate pedestrian trajectory by the relative posture of translation and rotation. Above all, this method does not require user's personal information, nor be limited to the mobile phone the way users carry. Therefore, not only both periodic and non-periodic can be used, but it can directly apply the six-dimensional data obtained by the inertial measurement units to reconstruct an accurate and reliable movement trajectory.

摘要 i
Abstract ii
誌謝 iii
目錄 iv
圖目錄 vi
表目錄 viii
第一章 緒論 1
1.1研究動機與目的 1
1.2研究方法 2
1.3論文架構 2
第二章 相關研究討論 3
2.1 三維空間之旋轉 3
2.1.1 四元數之基本定義 5
2.1.2 四元數之旋轉與插值 7
2.2 深度學習模型 8
2.2.1 長短期記憶網路 9
2.2.2 雙向長短期記憶網路(Bi-directional Long Short-Term Memory,Bi-LSTM) 10
2.2.3 自注意力機制 12
第三章 行人航位推算系統架構 14
3.1步伐檢測 14
3.2步長估計 17
3.3航向估測 17
第四章 系統架構 19
4.1校準與同步 20
4.1.1慣性里程數據集 20
4.1.2裝置座標系與世界座標系間轉換 21
4.1.3軌跡誤差 23
4.2原深度學習系統架構 24
4.3深度學習系統架構 25
4.3.1多任務學習(Multi-task Learning) 27
4.3.2模型之不確定性加權 28
第五章 實驗結果與分析 30
5.1相關說明 30
5.1.1數據集 30
5.1.2訓練與測試 30
5.1.3參數設置 32
5.2模擬結果一 32
5.3模擬結果二 42
5.4實驗分析 46
第六章 結論與未來展望 51
參考文獻 52

[1] R. Harle, "A Survey of Indoor Inertial Positioning Systems for Pedestrians", IEEE Communications Surveys & Tutorials, vol. 15, no. 3, pp. 1281-1293, 2013.
[2] P. Davidson, and R. Piché, "A Survey of Selected Indoor Positioning Methods for Smartphones", IEEE Communications Surveys & Tutorials, vol. 19, no. 2, pp. 1347-1370, 2017.
[3] P. Myung Chul, V. V. Chirakkal, and H. Dong Seog, "Robust pedestrian dead reckoning for indoor positioning using smartphone", IEEE International Conference on Consumer Electronics (ICCE), pp. 80-81, .2015.
[4] L. Hsu, Y. Gu, Y. Huang, and S. Kamijo, "Urban Pedestrian Navigation Using Smartphone-Based Dead Reckoning and 3-D Map-Aided GNSS", IEEE Sensors Journal, vol. 16, no. 5, pp. 1281-1293, 2016.
[5] C. Jiang, L. Xue, H. Chang, G. Yuan, and W. Yuan, "Signal Processing of MEMS Gyroscope
Arrays to Improve Accuracy Using a 1st Order Markov for Rate Signal Modeling", Sensors
(Basel, Switzerland), vol. 12, pp. 1720-37, 2012.
[6] B. Shin et al., "Motion Recognition-Based 3D Pedestrian Navigation System Using Smartphone", IEEE Sensors Journal, vol. 16, no. 18, pp. 6977-6989, 2016.
[7] R. Zhou, "Pedestrian dead reckoning on smartphones with varying walking speed", 2016 IEEE International Conference on Communications (ICC), pp. 1-6, 2016.
[8] P. Savage, "Strapdown Inertial Navigation Integration Algorithm Design Part 1: Attitude Algorithms", Journal of Guidance Control and Dynamics, vol. 21, pp. 19-28, 1998.
[9] T. Qin, P. Li, and S. Shen, "VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State Estimator", IEEE Transactions on Robotics, vol. 34, no. 4, pp. 1004-1020, 2018.
[10] R. Clark, S. Wang, H. Wen, A. Markham, and A. Trigoni, "VINet: Visual-Inertial Odometry as a Sequence-to-Sequence Learning Problem", in AAAI, 2017.
[11] S. Wang, R. Clark, H. Wen, and A. Trigoni, "DeepVO: Towards end-to-end visual odometry with deep Recurrent Convolutional Neural Networks", 2017 IEEE International Conference on Robotics and Automation (ICRA), pp. 2043-2050, 2017.
[12] C. Chen, C. X. Lu, J. Wahlström, A. Markham, and N. Trigoni, "Deep Neural Network Based Inertial Odometry Using Low-Cost Inertial Measurement Units", IEEE Transactions on Mobile Computing, vol. 20, no. 4, pp. 1351-1364, 2021.
[13] J. P. Silva do Monte Lima, H. Uchiyama, and R.-i. Taniguchi, "End-to-End Learning Framework for IMU-Based 6-DOF Odometry", Sensors, vol. 19, no. 17, 2019.
[14] D. E. Rumelhart, G. E. Hinton, and R. J. Williams, "Learning representations by back-propagating errors", Nature, vol. 323, no. 6088, pp. 533-536, 1986
[15] G. Hinton, "Learning multiple layers of representation", Trends in cognitive sciences, vol. 11, pp. 428-34, 2007.
[16] S. Hochreiter and J. Schmidhuber, "Long Short-Term Memory", Neural Computation, vol. 9, no. 8, pp. 1735-1780, 1997.
[17] Y. O. Ouma, R. Cheruyot, and A. N. Wachera, "Rainfall and runoff time-series trend analysis using LSTM recurrent neural network and wavelet neural network with satellite-based meteorological data: case study of Nzoia hydrologic basin", Complex & Intelligent Systems, 2021.
[18] M. Schuster and K. K. Paliwal, "Bidirectional recurrent neural networks", IEEE Transactions on Signal Processing, vol. 45, no. 11, pp. 2673-2681, 1997.
[19] M. Jokar and F. Semperlotti, "Finite Element Network Analysis: A Machine Learning based Computational Framework for the Simulation of Physical Systems", 2020.
[20] A. Graves and J. Schmidhuber, "Framewise phoneme classification with bidirectional LSTM and other neural network architectures", Neural Networks, vol. 18, no. 5, pp. 602-610, 2005.
[21] Z. Cui, R. Ke, Z. Pu, and Y. Wang, "Stacked Bidirectional and Unidirectional LSTM Recurrent Neural Network for Forecasting Network-wide Traffic State with Missing Values", ArXiv, vol. abs/2005.11627, 2020.
[22] V. Mnih, N. Heess, A. Graves, and K. Kavukcuoglu, "Recurrent Models of Visual Attention", in NIPS, 2014.
[23] W. Yin, S. Ebert, and H. Schütze, "Attention-Based Convolutional Neural Network for Machine Comprehension", ArXiv, vol. abs/1602.04341, 2016.
[24] A. Vaswani et al., "Attention is All you Need", ArXiv, vol. abs/1706.03762, 2017.
[25] Q. Tian, Z. Salcic, K. I. Wang, and Y. Pan, "A Multi-Mode Dead Reckoning System for Pedestrian Tracking Using Smartphones", IEEE Sensors Journal, vol. 16, no. 7, pp. 2079-2093, 2016.
[26] J. Hausdorff, "Gait dynamics, fractals and falls: Finding meaning in the stride-to-stride fluctuations of human walking", Human movement science, vol. 26, pp. 555-89, 2007.
[27] J. Scarlett, "Enhancing the performance of pedometers using a single accelerometer", 2007.
[28] C. Chen, P. Zhao, C. X. Lu, W. Wang, A. Markham, and A. Trigoni, "Deep-Learning-Based Pedestrian Inertial Navigation: Methods, Datfa Set, and On-Device Inference", IEEE Internet of Things Journal, vol. 7, pp. 4431-4441, 2020.
[29] J. Sturm, N. Engelhard, F. Endres, W. Burgard, and D. Cremers, "A benchmark for the evaluation of RGB-D SLAM systems", IEEE/RSJ International Conference on Intelligent Robots and Systems, 7-12 Oct. 2012 2012, pp. 573-580, 2012.
[30] Zhang and D. Scaramuzza, "A Tutorial on Quantitative Trajectory Evaluation for Visual(-Inertial) Odometry", in 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 7244-7251, 2018.
[31] S. Ruder, "An Overview of Multi-Task Learning in Deep Neural Networks", ArXiv, vol. abs/1706.05098, 2017.
[32] A. Kendall, Y. Gal, and R. Cipolla, "Multi-task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics", IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 7482-7491, 2018.