簡易檢索 / 詳目顯示

回結果列表
研究生： 戴勵揚
Tai, Li-Yang
論文名稱： 基於深度學習的行動外骨骼輔具失衡與跌倒偵測
Balance Loss and Fall Detection for Mobility Assist Exoskeletons Based on Deep Learning
指導教授： 黃致憲
Huang, Chih-Hsien
學位類別： 碩士
Master
系所名稱： 電機資訊學院 - 電機工程學系
Department of Electrical Engineering
論文出版年： 2024
畢業學年度： 113
語文別： 英文
論文頁數： 73
中文關鍵詞： 行動輔具 、跌倒偵測 、失衡偵測 、深度學習 、閾值法
外文關鍵詞： Assistive Device, Fall Detection, Balance Loss Detection, Deep Learning, Threshold Method
相關次數： 點閱：118  下載：8 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • 高齡化社會中，隨著年齡增長，許多人逐漸面臨運動功能退化的挑戰，而行走能力的喪失更是加劇了這個問題，對整體健康帶來連鎖影響。行動輔助外骨骼可以有效協助行動不便的使用者維持或恢復行走能力。然而，外骨骼的效能取決於其能否準確掌握使用者的狀況，尤其是當使用者失去行走能力的情境，這些關鍵時刻的檢測至關重要。若能準確辨識這些情況，不僅能顯著提升外骨骼的功能，也能大幅改善使用者的安全性。
    行走能力的喪失通常與跌倒有關，而跌倒可能由絆倒、滑倒、昏厥或外骨骼故障等原因引起。其中，絆倒或滑倒可能導致使用者撞擊周遭物體或地面，造成潛在的受傷風險。因此，撞擊前預測跌倒至關重要，其主要特徵為失去平衡。本研究採用深度神經網路技術，透過分析時間序列數據來判斷是否失去平衡。
    至於由昏厥或肌力不足引起的跌倒，可能呈現緩慢下降的模式，這類跌倒可能無法透過傳統加速度平衡檢測法準確辨識。然而，由於緩慢跌倒通常不會造成嚴重撞擊，因此更適合使用基於閾值的檢測方法，例如透過腳底壓力感測器進行判斷，這種方法強調準確性而非檢測速度。結合深度學習的失衡偵測與基於閾值的跌倒偵測，可以大幅提升運動輔助外骨骼對使用情境的理解，進一步增強其安全性與實用性，讓使用者在行動輔助上獲得保障。

    An aging population often faces declining motor function, and the loss of walking ability exacerbates this issue, leading to a cascading effect on overall health. Movement assist exoskeleton can help mobility-impaired users maintain or regain their walking capabilities. The effectiveness of exoskeleton relies on its ability to interpret context accurately, and situations where users lose locomotion are crucial to detect. Successfully detecting these occurrences can significantly improve the function of movement assist exoskeleton. Loss of locomotion usually occurs due to a fall, which may result from tripping, slipping, fainting, or exoskeleton malfunction. In cases such as tripping and slipping, the user may impact nearby objects or the floor, potentially causing injury. Pre-impact fall detection is essential in these scenarios, with the defining characteristic of pre-impact being a loss of balance. This study uses deep neural network to analyze time series data and classify whether a balance loss is occurring. Falls due to fainting or loss of strength may involve a more gradual descent, potentially bypassing acceleration-based balance loss detection. Since slow falls typically do not result in severe impact, a threshold-based fall detection method, utilizing pressure sensors under the feet, which emphasizes accuracy more than detection speed, is more suitable. By combining deep neural network-based balance loss detection and threshold-based fall detection, the context received by movement assist exoskeleton is significantly improved, enhancing both safety and effectiveness.

    摘要 III ABSTRACT IV ACKNOWLEDGEMENTS V CONTENTS VI LIST OF TABLES VIII LIST OF FIGURES IX INTRODUCTION 1 1-1 Research Background 1 1-2 Related Work 2 1-3 Research Objectives 4 1-4 Thesis Structure 4 MATERIAL AND PRINCIPLE 7 2-1 Introduction 7 2-2 Mobility Assist Exoskeleton 7 2-3 Movement Detection System 8 2-4 Sensor Specifications 9 2-4-1 Three-axis gyroscope and three-axis accelerometer sensor (MPU-6050) 9 2-4-2 Force Sensing Resistor (FSR402) 11 2-5 Raspberry Pi 4 Model B 12 2-6 Balance Loss and Fall Phase 14 2-7 Deep Neural Network Training and Utilization 15 2-7-1 Convolutional Neural Network 16 2-7-3 Max Pooling 17 2-7-2 Long Short-Term Memory 17 2-7-3 Dense 19 SYSTEM DESIGN AND METHODS 20 3-1 Sensor Signal Processing 20 3-2 Sensor Signal Preprocessing 20 3-2-1 MPU6050 Signal Preprocessing 20 3-2-2 Foot Pressure Sensor Signal Preprocessing 21 3-3 Data Preprocessing 23 3-3-1 Sum Vector Magnitude 24 3-3-2 Tilt Angle 24 3-3-3 Sliding Window 25 3-3-4 Data Normalization 26 3-4 Balance Loss and Fall Detection 27 3-4-1 Balance Loss vs Fall Detection 28 3-4-2 Balance Loss Detection 29 3-4-3 Fall Detection 29 EXPERIMENT AND RESULTS 31 4-1 Experiment Purpose 31 4-2 Experiment Structure 31 4-3 Data Acquisition Device Hardware 33 4-4 Data Acquisition 34 4-5 Deep Learning Model Structure and Training 39 4-5-1 Threshold Application 42 4-6 Experiment Result and Analysis 46 4-6-1 Choice of Metrics 46 4-6-2 Balance Loss Detection Result 48 4-6-3 Fall Detection Result 52 4-6-4 Balance Loss and Fall Detection Result 54 4-7 Summary and Discussion 55 CONCLUSION AND FUTURE WORK 57 5-1 Conclusion 57 5-2 Future Work 58 REFERENCE 60

    [1] R. Cuevas-Trisan, "Balance Problems and Fall Risks in the Elderly," in Clinics in Geriatric Medicine, vol. 35, no. 2, pp. 173-183, Apr. 2019
    [2] M. R. Kosorok, G. S. Omenn, P. Diehr, T. D. Koepsell, and D. L. Patrick, "Restricted activity days among older adults," in American Journal of Public Health, vol. 82, no. 9, pp. 1263-1267, Sep. 1992.
    [3] T. Burton, É. Séguin and M. Doumit, "Study of Hip Exoskeleton Technology for Elderly Stability During Walking," 2022 World Automation Congress (WAC), San Antonio, TX, USA, 2022, pp. 596-601
    [4] T. Seene and P. Kaasik, "Muscle weakness in the elderly: Role of sarcopenia, dynapenia, and possibilities for rehabilitation," in European Review of Aging and Physical Activity, vol. 9, pp. 109-117, 2012.
    [5] M. Vallejo, C. V. Isaza and J. D. López, "Artificial Neural Networks as an alternative to traditional fall detection methods," 2013 35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Osaka, Japan, 2013, pp. 1648-1651
    [6] M. C. Nevitt, S. R. Cummings, and E. S. Hudes, "Risk Factors for Injurious Falls: a Prospective Study," in Journal of Gerontology, vol. 46, no. 5, pp. M164-M170, Sep. 1991
    [7] L. Z. Rubenstein, "Falls in older people: epidemiology, risk factors and strategies for prevention," in Age and Ageing, vol. 35, no. suppl_2, pp. ii37-ii41, Sep. 2006
    [8] C. -P. Liu et al., "Deep Learning-based Fall Detection Algorithm Using Ensemble Model of Coarse-fine CNN and GRU Networks," 2023 IEEE International Symposium on Medical Measurements and Applications (MeMeA), pp. 1-5, 2023
    [9] A. A. Zecevic, A. W. Salmoni, M. Speechley, and A. A. Vandervoort, "Defining a fall and reasons for falling: Comparisons among the views of seniors, health care providers, and the research literature," in The Gerontologist, vol. 46, no. 3, pp. 367-376, Jun. 2006
    [10] M. N. Nyan, F. E. H. Tay, and E. Murugasu, "A wearable system for pre-impact fall detection," in Journal of Biomechanics, vol. 41, no. 16, pp. 3475-3481, 2008
    [11] M. Tolkiehn, L. Atallah, B. Lo and G. -Z. Yang, "Direction sensitive fall detection using a triaxial accelerometer and a barometric pressure sensor," 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Boston, MA, USA, 2011, pp. 369-372
    [12] X. Yu, B. Koo, J. Jang, Y. Kim, and S. Xiong, "A comprehensive comparison of accuracy and practicality of different types of algorithms for pre-impact fall detection using both young and old adults," in Measurement, vol. 201, pp. 111785, 2022
    [13] T. H. Kim, A. Choi, H. M. Heo, K. Kim, K. Lee, and J. H. Mun, "Machine Learning-Based Pre-Impact Fall Detection Model to Discriminate Various Types of Fall," in Journal of Biomechanical Engineering, vol. 141, no. 8, pp. 081010, Aug. 2019
    [14] J. Shi, D. Chen, and M. Wang, "Pre-Impact Fall Detection with CNN-Based Class Activation Mapping Method," in Sensors, vol. 20, no. 17, pp. 4750, 2020
    [15] O. Aziz, C. M. Russell, E. J. Park and S. N. Robinovitch, "The effect of window size and lead time on pre-impact fall detection accuracy using support vector machine analysis of waist mounted inertial sensor data," 2014 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Chicago, IL, USA, 2014, pp. 30-33
    [16] F. A. S. Ferreira de Sousa, C. Escriba, E. G. Avina Bravo, V. Brossa, J. -Y. Fourniols and C. Rossi, "Wearable Pre-Impact Fall Detection System Based on 3D Accelerometer and Subject’s Height," in IEEE Sensors Journal, vol. 22, no. 2, pp. 1738-1745, Jan. 15, 2022
    [17] A. Sucerquia, J. D. López, and J. F. Vargas-Bonilla, "SisFall: A Fall and Movement Dataset," in Sensors, vol. 17, no. 1, pp. 198
    [18] T. Tamura, T. Yoshimura, M. Sekine, M. Uchida and O. Tanaka, "A Wearable Airbag to Prevent Fall Injuries," in IEEE Transactions on Information Technology in Biomedicine, vol. 13, no. 6, pp. 910-914, Nov. 2009
    [19] G. Shi, C. S. Chan, W. J. Li, K. -S. Leung, Y. Zou and Y. Jin, "Mobile Human Airbag System for Fall Protection Using MEMS Sensors and Embedded SVM Classifier," in IEEE Sensors Journal, vol. 9, no. 5, pp. 495-503, May 2009
    [20] V. B. Semwal, S. A. Katiyar, R. Chakraborty, and G. C. Nandi, "Biologically-inspired push recovery capable bipedal locomotion modeling through hybrid automata," in Robotics and Autonomous Systems, vol. 70, pp. 181-190, 2015
    [21] X. Yu, J. Wan, G. An, X. Yin, and S. Xiong, "A novel semi-supervised model for pre-impact fall detection with limited fall data," in Engineering Applications of Artificial Intelligence, vol. 132, pp. 108469, 2024
    [22] R. Jain and V. B. Semwal, "A Novel Feature Extraction Method for Preimpact Fall Detection System Using Deep Learning and Wearable Sensors," in IEEE Sensors Journal, vol. 22, no. 23, pp. 22943-22951, Dec. 1, 2022
    [23] X. Yu, J. Jang, and S. Xiong, "A Large-Scale Open Motion Dataset (KFall) and Benchmark Algorithms for Detecting Pre-impact Fall of the Elderly Using Wearable Inertial Sensors," in Frontiers in Aging Neuroscience, vol. 13, 2021
    [24] A. Benoit, C. Escriba, D. Gauchard, A. Esteve and C. Rossi, "Analyzing and Comparing Deep Learning Models on an ARM 32 Bits Microcontroller for Pre-Impact Fall Detection," in IEEE Sensors Journal, vol. 24, no. 7, pp. 11829-11842, April 1, 2024
    [25] X. Chen, S. Jiang and B. Lo, "Subject-Independent Slow Fall Detection with Wearable Sensors via Deep Learning," 2020 IEEE SENSORS, pp. 1-4, 2020
    [26] A. T. Özdemir, "An analysis on sensor locations of the human body for wearable fall detection devices: Principles and practice," in Sensors, vol. 16, no. 8, pp. 1161, 2016
    [27] K. -C. Liu, C. -Y. Hsieh, S. J. -P. Hsu and C. -T. Chan, "Impact of Sampling Rate on Wearable-Based Fall Detection Systems Based on Machine Learning Models," in IEEE Sensors Journal, vol. 18, no. 23, pp. 9882-9890, Dec. 1, 2018
    [28] M. -K. Yi, K. Han and S. O. Hwang, "Fall Detection of the Elderly Using Denoising LSTM-Based Convolutional Variant Autoencoder," in IEEE Sensors Journal, vol. 24, no. 11, pp. 18556-18567, June 1, 2024
    [29] K. -C. Liu, C. -Y. Hsieh, H. -Y. Huang, S. J. -P. Hsu and C. -T. Chan, "An Analysis of Segmentation Approaches and Window Sizes in Wearable-Based Critical Fall Detection Systems With Machine Learning Models," in IEEE Sensors Journal, vol. 20, no. 6, pp. 3303-3313, March 15, 2020
    [30] M. Saleh, M. Abbas and R. B. Le Jeannès, "FallAllD: An Open Dataset of Human Falls and Activities of Daily Living for Classical and Deep Learning Applications," in IEEE Sensors Journal, vol. 21, no. 2, pp. 1849-1858, Jan. 15, 2021

    下載圖示 校內：立即公開
    校外：立即公開
    QR CODE