步態分析之研究已普遍運用於許多領域之中，如運動生物力學、醫療診斷與傷害預防等。其中足部壓力動態表現參數，特別是地面反作用力之垂之分量(vGRF)及壓力中心(CoP)的變化軌跡可顯示出人體運動狀態與平衡能力等相關資訊，目前在量測步態資訊主要使用以下三種方法：影像辨識、固定式感測器、穿戴式感測器，影像辨識與固定式感測器擁有較高的精準度，但其缺點為設備價格昂貴且量測場域受限；穿戴式感測器擁有便攜性與價格較低之優點，然而接觸型感測元件其量測精準度隨使用時間降低，需要定期汰換。
在本文中，我們提出了一個使用6軸慣性量測單元(IMU)估算vGRF及CoP軌跡之方法，IMU常用來量測運動時的姿態與位置偏移，由於IMU為非接觸式感測器，相較於接觸式壓力感測更不易因形變而影響其精準度。在所設計的實驗中，我們使用4個IMU分別置於左右腳、左小腿與腰前，並以商用壓力感測系統F-scan作為真實參考值。
於資料分析中，我們使用加速度與角速度資訊，基於機器學習方法中之長短期記憶模型預測vGRF及CoP軌跡，亦將於內文中說明相對應的軟硬體架構、資料格式、與機器學習方式；實驗中包含33,880個取樣點，其中70%訓練，其餘30%測試，初步實驗顯示，預測vGRF峰值之均方根誤差=4.83N，相當4.025%，預測CoP軌跡與實際軌跡之壓力中心偏移誤差為0.14公分。

Gait analysis has become prevalent in many fields such as sports biomechanics, medical diagnostics, and injury prevention. For the plantar pressure dynamic estimation in gait analysis, the vertical component of ground reaction force (vGRF) and center of pressure (CoP) trajectories are vital parameters about human locomotion and balance.

Three main approaches are used in measuring the gait parameters, namely computer vision, floor sensors, and wearables. Though the first two techniques are accurate, they are expensive and limited in the sensing area. The wearable devices have advantages in portable and low cost. However, the contact sensing like pressure detector suffers from long-term reliability.

In this research, we employ the 6DOF inertial measurement unit (IMU) (gyro system, Taiwan) to estimate the vGRF and CoP trajectories. Unlike conventional pressure sensing, IMUs are low cost and durable. Four IMUs are attached on the heel of two feet, left shank, and waist. F-scan pressure sensing system serves as the ground truth. Associated with hardware setup, we propose a LSTM model to predict vGRF and CoP, based on the acceleration and angle velocity data from IMUs. The data synchronization, data formation, and LSTM structure are explained. Under 33,880 sample points of normal walks, the first 70% is for training and the last 30% is for testing. Experiment shows the root mean square error of peak value of vGRF is equal to 4.83N, the error is 4.025% and the root mean square error of CoP excursion is 0.14cm.

摘要 II
ABSTRACT IV
LIST OF FIGURES VI
LIST OF TABLES VIII

CHAPTER 1 1

CHAPTER 2 4
2.1 GAIT KINEMATIC CHARACTERISTICS 4
2.2 TOOLS FOR MEASUREMENT 6
2.2.1 Computer Visions 6
2.2.2 Floor Sensors 7
2.2.3 Wearable Devices 8
2.3 INERTIAL MEASUREMENT UNIT (IMU) 10
2.4 COMPARISONS BETWEEN DIFFERENT TOOLS 11

CHAPTER 3 12
3.1 MACHINE LEARNING IN TIME-SERIES PREDICTION 12
3.2 VANILLA RNN 13
3.3 LONG SHORT-TERM MEMORY (LSTM) 15

CHAPTER 4 17
4.1 EXPERIMENT SETUP 17
4.1.1 Hardware systems 17
4.1.2 Experiment Setup 19
4.2 DATA PRE-PROCESSING 20
4.2.1 Synchronization between sensors 20
4.2.2 Feature Selection 23
4.2.3 Data Formation 25

CHAPTER 5 27
5.1 STRUCTURE OF THE PROPOSED LSTM 27
5.2 EXPERIMENT RESULTS FOR COP COORDINATES 30
5.3 EXPERIMENT RESULTS FOR VGRF 32

CHAPTER 6 34
6.1 PREDICTED ERROR OF COP ANALYSIS 34
6.2 PREDICTED ERROR OF VGRF ANALYSIS 36

REFERENCE 37

[1] T. R. Han, N. J. Paik, M. S., "Quantification of the path of center of pressure (COP) using an F-scan in-shoe transducer," Im Gait Posture, 1999 Dec;10(3): 248–254

[2] H. Sobral et al., "Human gait analysis using instrumented shoes," 2015 IEEE 4th Portuguese Meeting on Bioengineering (ENBENG), Porto, 2015, pp. 1-1.
doi: 10.1109/ENBENG.2015.7088807

[3] Headon, Robert and Rupert Curwen, "Recognizing movements from the ground reaction force," PUI '01, 2001.

[4] Joseph P. Hunter, Robert N. Marshall, Peter J. McNair, "Relationships between ground reaction force impulse and kinematics of sprint-running acceleration," J Appl Biomech, 2005 Feb; 21(1): 31–43.

[5] Jan Rueterbories, Erika G. Spaich, Birgit Larsen, Ole K. Andersen, "Methods for gait event detection and analysis in ambulatory systems," Med Eng Phys, 2010 Jul; 32(6): 545–552. doi: 10.1016/j.medengphy.2010.03.007

[6] Farahpour, Nader et al, "Gait ground reaction force characteristics of low back pain patients with pronated foot and able-bodied individuals with and without foot pronation," Journal of Biomechanics, vol. 49, Issue 9, 1705 – 1710, 2016.

[7] Alexander Ruhe, René Fejer, Bruce Walker, "Center of pressure excursion as a measure of balance performance in patients with non-specific low back pain compared to healthy controls: a systematic review of the literature," Eur Spine J, 2011 Mar; 20(3): 358–368. Published online 2010 Aug 19. doi: 10.1007/s00586-010-1543-2

[8] Chris J. Hass, Dwight E. Waddell, Richard P. Fleming, Jorge L. Juncos, Robert J. Gregor, "Gait initiation and dynamic balance control in Parkinson's disease," Arch Phys Med Rehabil, 2005 Nov; 86(11): 2172–2176. doi: 10.1016/j.apmr.2005.05.013

[9] Razak AHA, Zayegh A, Begg RK, Wahab Y, "Foot Plantar Pressure Measurement System: A Review," Sensors (Basel, Switzerland), 2012;12(7):9884-9912. doi:10.3390/s120709884.

[10] Susan Hillier, Myra S. Lai, "Insole Plantar Pressure Measurement During Quiet Stance Post Stroke," Top Stroke Rehabil, 2009 May-Jun; 16(3): 189–195. doi: 10.1310/tsr1603-189

[11] H. M. Schepers*, E. H. F. van Asseldonk, J. H. Buurke and P. H. Veltink, "Ambulatory Estimation of Center of Mass Displacement During Walking," in IEEE Transactions on Biomedical Engineering, vol. 56, no. 4, pp. 1189-1195, April 2009.
doi: 10.1109/TBME.2008.2011059

[12] T. Maudie and J. Wertz, "Pressure sensor performance and reliability," in IEEE Industry Applications Magazine, vol. 3, no. 3, pp. 37-43, May-June 1997.
doi: 10.1109/2943.589894

[13] H. M. Schepers, H. F. J. M. Koopman and P. H. Veltink, "Ambulatory Assessment of Ankle and Foot Dynamics," in IEEE Transactions on Biomedical Engineering, vol. 54, no. 5, pp. 895-902, May 2007.
doi: 10.1109/TBME.2006.889769

[14] Zachary C. Lipton, John Berkowitz, Charles Elkan, "A Critical Review of Recurrent Neural Networks for Sequence Learning," arXiv preprint arXiv:1506.00019, 2015

[15] S. Hochreiter, "The vanishing gradient problem during learning recurrent neural nets and problem solutions," International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, vol. 6, no. 02, pp. 107–116, 1998

[16] O. Vinyals, A. Toshev, S. Bengio and D. Erhan, "Show and tell: A neural image caption generator," 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, MA, 2015, pp. 3156-3164.
doi: 10.1109/CVPR.2015.7298935

[17] S. Hochreiter, J. Schmidhuber, "Long short-term memory," Neural Comput. 1997 Nov 15; 9(8): 1735–1780

[18] Hu X, Zhao J, Peng D, Sun Z, Qu X, "Estimation of Foot Plantar Center of Pressure Trajectories with Low-Cost Instrumented Insoles Using an Individual-Specific Nonlinear Model," Sensors (Basel, Switzerland). 2018;18(2):421. doi:10.3390/s18020421

[19] Park S, Ko Y-M, Park J-W, "The Correlation between Dynamic Balance Measures and Stance Sub-phase COP Displacement Time in Older Adults during Obstacle Crossing," Journal of Physical Therapy Science, 2013;25(9):1193-1196. doi:10.1589/jpts.25.1193

[20] Henk van der Worp, Jelte W. Vrielink, Steef W. Bredeweg, "Do runners who suffer injuries have higher vertical ground reaction forces than those who remain injury-free? A systematic review and meta-analysis," Br J Sports Med. 2016 Apr; 50(8): 450–457, Published online 2016 Jan 4. doi: 10.1136/bjsports-2015-094924