研究生: |
吳敦晏 DUN-YAN WU |
---|---|
論文名稱: |
基於循環神經網路模型之穿戴動力外骨骼輔具爬梯膝關節扭矩估測研究 RNN based Knee Joint Muscular Torque Estimation of Wearable Powered Exoskeleton for Stair Climbing Applications |
指導教授: |
林紀穎
Chi-Ying Lin |
口試委員: |
黃緒哲
Shiuh-Jer Huang 劉孟昆 Meng-Kun Liu |
學位類別: |
碩士 Master |
系所名稱: |
工程學院 - 機械工程系 Department of Mechanical Engineering |
論文出版年: | 2020 |
畢業學年度: | 108 |
語文別: | 中文 |
論文頁數: | 141 |
中文關鍵詞: | 肌少症 、上樓梯 、外骨骼輔具 、膝關節扭矩估測 、循環神經網路模型 |
外文關鍵詞: | Sarcopenia, Stair-climbing, Exoskeleton, Knee joint torque estimation, Recurrent neural network |
相關次數: | 點閱:213 下載:0 |
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 |
在自動化技術日漸普及的現代社會中,「肌少症」愈發容易出現於年齡增長的人們,甚至營養不均衡與少活動的年輕人也是值得注意的潛在高風險族群。然而「樓梯」這個場景仍然是臺灣日常生活最常見的路況之一,因此這類看似健康的人群在爬樓梯的過程中可能會增加心血管負擔進而引發相關健康問題;如此除了無法持續保持正常的爬梯步態,亦會大幅增加在樓梯間跌落的風險。在爬梯的過程中膝關節為最費力的身體部位,為此本研究針對上樓梯的應用開發能夠於適當時機給予右膝關節輔助力的外骨骼輔具。我們首先推導人體下肢爬梯的動態模型與量測人體下肢姿態以及腳底力量分佈,藉此分析爬梯步態行為以及估測膝關節於此過程中所需扭矩變化。然而上述估測扭矩的方法需在腳底裝設力量感測器,在實際應用中具有輔具穿戴不易、與行走不便等諸多缺點。由於爬梯是一個具連貫性動作,因此非常適合利用循環神經網路模型(RNN)找出下肢姿態與膝關節扭矩之間的關係。藉由訓練後的類神經網路模型,外骨骼輔具將可直接透過下肢姿態精準地給予適當的爬梯輔助力;此法具有不需依據步態階段切換控制法則的優點,並省去腳底力量感測器裝設的費用與不便,將可提高外骨骼輔具爬梯舒適性與安全性。本研究將穿著外骨骼實際爬梯實驗分為三種情形進行探討,分別為(1)無人機互動與輔助力;(2)有人機互動且輔助力參數 為0;(3)有人機互動且輔助力參數 為4。以上實驗主要探討在有無存在人機互動下對RNN輸出的影響、以及實際量測外骨骼給予輔助力時下肢肌肉活動量的變化。結果顯示不論是否存在人機互動,皆不會影響RNN的輸出並可準確判斷施力時機點,亦即所訓練的RNN具有不錯的強健性。外骨骼給予輔助力初步實驗結果顯示下肢肌肉活動量並無明顯下降的趨勢,仍然有待後續進一步外骨骼系統改良修正並進行較為長期的臨床穿戴驗證。
Stair-climbing is a laborious task for people with Sarcopenia, in which these healthy-seemly people may have a relatively higher chance of falling from the stairs than healthy people. Referring to the fact that knee joints contribute most during stair-climbing, this thesis proposes an exoskeleton assistive device to provide knee joint assistance. In this study, we derive the dynamic model of lower limbs during stair-climbing and measure the posture of the lower limbs as well as the force distribution of the feet. Using this information, we analyze the gait behavior and particularly estimate the torque of knee joint during stair-climbing. Because installation of force sensor on feet causes users great inconvenience in the wearing & walking process and most importantly, the fact that stair-climbing is a sequential locomotive behavior makes the Recurrent Neural Network Model (RNN) an excellent tool to derive the relationship between the posture of the lower limbs and the torque of the knee joint. Thanks to the use of RNN, there is no need for exoskeleton system to switch between several discrete high level control algorithms developed in each state of stair-climbing and the system is meant to offer appropriate assistance by just using the posture of the lower limbs, i.e., no force sensors are required in our developed system anymore. In this study, three kinds of stair-climbing experiments are conducted to investigate the effects of the RNN output with or without HRI and evaluate the muscle activities from human lower limbs especially when the exoskeleton provides assistance to human, including (1) no human robot interaction (HRI) and zero assistance; (2) with HRI and assistive factor γ = 0; (3) with HRI and assistive factor γ = 4. The results show that the existence of HRI minimally affects the output of the RNN which assures the proper timing of assistance generation, meaning that the trained RNN model has pretty good robustness. However, the measurement results from EMG sensors indicate that there is little variation in the muscle activities of subject’s lower limbs. Further modifications of the exoskeleton mechatronic system and clinical performance evaluation methods should be carefully investigated with a well-designed long term use protocol in the future study.
[1] V. Santilli, A. Bernetti, M. Mangone, and M. Paoloni, "Clinical definition of sarcopenia," (in eng), Clinical cases in mineral and bone metabolism : the official journal of the Italian Society of Osteoporosis, Mineral Metabolism, and Skeletal Diseases, vol. 11, no. 3, pp. 177-180, 2014.
[2] 室外台階圖, 檢自:https://mts.jk51.com/tushuo/4628087.html.
[3] 防止意外樓梯安全關鍵在「高度」。, 檢自:https://news.housefun.com.tw/news/article/185374171661.html.
[4] Z. Zhang et al., "Lower Extremity Exoskeleton for Stair Climbing Augmentation," in 2018 3rd International Conference on Advanced Robotics and Mechatronics (ICARM), 2018, pp. 762-768.
[5] 吳健平, "可變剛性致動器開發與飛躍抓枝機器人著陸效能改良之應用," 國立台灣科技大學機械系碩士學位論文, 2019.
[6] S. Kim and J. Bae, "Force-Mode Control of Rotary Series Elastic Actuators in a Lower Extremity Exoskeleton Using Model-Inverse Time Delay Control," IEEE/ASME Transactions on Mechatronics, vol. 22, no. 3, pp. 1392-1400, 2017.
[7] F. Patané, S. Rossi, F. D. Sette, J. Taborri, and P. Cappa, "WAKE-Up Exoskeleton to Assist Children With Cerebral Palsy: Design and Preliminary Evaluation in Level Walking," IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 25, no. 7, pp. 906-916, 2017.
[8] M. K. Shepherd and E. J. Rouse, "Design and Validation of a Torque-Controllable Knee Exoskeleton for Sit-to-Stand Assistance," IEEE/ASME Transactions on Mechatronics, vol. 22, no. 4, pp. 1695-1704, 2017.
[9] S. Kim and J. Bae, "Development of a lower extremity exoskeleton system for human-robot interaction," in 2014 11th International Conference on Ubiquitous Robots and Ambient Intelligence (URAI), 2014, pp. 132-135.
[10] S. Wang et al., "Design and Control of the MINDWALKER Exoskeleton," IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 23, 10/30 2014.
[11] R. Van Ham, B. Vanderborght, M. Van Damme, B. Verrelst, and D. Lefeber, "MACCEPA, the mechanically adjustable compliance and controllable equilibrium position actuator: Design and implementation in a biped robot," Robotics and Autonomous Systems, vol. 55, no. 10, pp. 761-768, 2007/10/31/ 2007.
[12] V. Grosu, C. Rodriguez-Guerrero, S. Grosu, B. Vanderborght, and D. Lefeber, "Design of Smart Modular Variable Stiffness Actuators for Robotic-Assistive Devices," IEEE/ASME Transactions on Mechatronics, vol. 22, no. 4, pp. 1777-1785, 2017.
[13] R. Furnémont, G. Mathijssen, T. Verstraten, D. Lefeber, and B. Vanderborght, "Bi-directional series-parallel elastic actuator and overlap of the actuation layers," Bioinspiration & Biomimetics, vol. 11, 12/26 2016.
[14] K. A. Witte, A. M. Fatschel, and S. H. Collins, "Design of a lightweight, tethered, torque-controlled knee exoskeleton," in 2017 International Conference on Rehabilitation Robotics (ICORR), 2017, pp. 1646-1653.
[15] P. Joudzadeh, A. Hadi, B. Tarvirdizadeh, D. Borooghani, and K. Alipour, "Design and fabrication of a lower limb exoskeleton to assist in stair ascending," Industrial Robot: the international journal of robotics research and application, vol. 46, no. 2, pp. 290-299, 2019.
[16] J. Wang et al., "Comfort-Centered Design of a Lightweight and Backdrivable Knee Exoskeleton," IEEE Robotics and Automation Letters, vol. 3, no. 4, pp. 4265-4272, 2018.
[17] B. Celebi, M. Yalcin, and V. Patoglu, "AssistOn-Knee: A self-aligning knee exoskeleton," in 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2013, pp. 996-1002.
[18] X. Liu and Q. Wang, "Real-Time Locomotion Mode Recognition and Assistive Torque Control for Unilateral Knee Exoskeleton on Different Terrains," IEEE/ASME Transactions on Mechatronics, pp. 1-1, 2020.
[19] D. Xu, X. Liu, and Q. Wang, "Knee Exoskeleton Assistive Torque Control Based on Real-Time Gait Event Detection," IEEE Transactions on Medical Robotics and Bionics, vol. 1, no. 3, pp. 158-168, 2019.
[20] D. X. Phu, T. D. Huy, and T. H. Ha, "An Implementation of Neuro-Fuzzy System for Gait Analysis in a Smart Insole of Exoskeleton," in 2018 4th International Conference on Green Technology and Sustainable Development (GTSD), 2018, pp. 611-614.
[21] Susanto, R. Analia, and K. Song, "Design of assistive torque for a lower limb exoskeleton based on motion prediction," in 2016 55th Annual Conference of the Society of Instrument and Control Engineers of Japan (SICE), 2016, pp. 172-177.
[22] I. Kang, P. Kunapuli, and A. J. Young, "Real-Time Neural Network-Based Gait Phase Estimation Using a Robotic Hip Exoskeleton," IEEE Transactions on Medical Robotics and Bionics, vol. 2, no. 1, pp. 28-37, 2020.
[23] D. Liu, W. Du, X. Wu, C. Wang, and Y. Qiao, "Deep rehabilitation gait learning for modeling knee joints of lower-limb exoskeleton," in 2016 IEEE International Conference on Robotics and Biomimetics (ROBIO), 2016, pp. 1058-1063.
[24] L. Fengyan et al., Online Adaptive and LSTM-Based Trajectory Generation of Lower Limb Exoskeletons for Stroke Rehabilitation. 2018, pp. 27-32.
[25] P. T. Chinimilli, Z. Qiao, S. M. Rezayat Sorkhabadi, V. Jhawar, I. H. Fong, and W. Zhang, "Automatic virtual impedance adaptation of a knee exoskeleton for personalized walking assistance," Robotics and Autonomous Systems, vol. 114, pp. 66-76, 2019/04/01/ 2019.
[26] K. Seo et al., "RNN-Based On-Line Continuous Gait Phase Estimation from Shank-Mounted IMUs to Control Ankle Exoskeletons," in 2019 IEEE 16th International Conference on Rehabilitation Robotics (ICORR), 2019, pp. 809-815.
[27] A. M. Boudali, P. J. Sinclair, and I. R. Manchester, "Predicting Transitioning Walking Gaits: Hip and Knee Joint Trajectories From the Motion of Walking Canes," IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 27, no. 9, pp. 1791-1800, 2019.
[28] J. C. Pérez-Ibarra, H. Williams, A. A. G. Siqueira, and H. I. Krebs, "Real-Time Identification of Impaired Gait Phases Using a Single Foot-Mounted Inertial Sensor: Review and Feasibility Study," in 2018 7th IEEE International Conference on Biomedical Robotics and Biomechatronics (Biorob), 2018, pp. 1157-1162.
[29] "Discovery kit with STM32F407VG MCU - STMicroelectronics," 檢自:https://www.st.com/resource/en/user_manual/dm00039084-discovery-kit-with-stm32f407vg-mcu-stmicroelectronics.pdf.
[30] "Arduino pro mini," 檢自:https://www.taiwaniot.com.tw/product/arduino-pro-mini/.
[31] "MPU6050," 檢自:https://www.jsumo.com/mpu6050-gy521-cheap-6dof-imu-accelerometer-gyro.
[32] "MPU6050 Datasheet," 檢自:https://invensense.tdk.com/wp-content/uploads/2015/02/MPU-6000-Datasheet1.pdf.
[33] "TCA9548A – I2C multiplexer," 檢自:http://domoticx.com/module-tca9548a-i2c-multiplexer/.
[34] "Tekscan FSR-A301," 檢自:https://www.pcstore.com.tw/mems/M30221610.htm.
[35] "ADS1115 A/D轉換器," 檢自:http://850121.blogspot.com/2018/12/arduino-four-ads1115-modules.html.
[36] "ADS1115 A/D Datasheet," 檢自:https://www.ti.com/lit/ds/symlink/ads1115.pdf?ts=1597115819714.
[37] "北河有刷馬達3890系列 Datasheet," 檢自:http://www.nrc.com.tw/CAT/Faulhaber/3863_CR.pdf.
[38] "三柱蝸桿蝸輪 Datasheet," 檢自:https://tw.misumi-ec.com/pdf/fa/2019/p1263.pdf.
[39] "鋰聚合物電池 Datasheet," 檢自:https://www.ruten.com.tw/item/show?21910124910325.
[40] "大功率降壓模組 Datasheet," 檢自:https://www.ruten.com.tw/item/show?21604763214646.
[41] "NodeMCU-32S微處理機," 檢自:https://www.taiwansensor.com.tw/product/nodemcu-32s-lua-%E7%89%A9%E8%81%AF%E7%B6%B2%E9%96%8B%E7%99%BC%E6%9D%BF-wifi%E8%97%8D%E7%89%992%E5%90%881%E9%9B%99%E6%A0%B8cpu-%E4%BD%8E%E5%8A%9F%E8%80%97-%E5%9F%BA%E6%96%BCesp32/.
[42] "NodeMCU-32S Datasheet," 檢自:http://wiki.ai-thinker.com/_media/esp32/docs/nodemcu-32s_product_specification.pdf.
[43] "HX711 Datasheet," 檢自:https://cdn.sparkfun.com/datasheets/Sensors/ForceFlex/hx711_english.pdf.
[44] Q. Wu, X. Wang, X. Zhang, B. Chen, Z. Shao, and L. Lu, "Development of a wearable multi-sensor system for lower limb joint torque evaluation during stairs climbing," in 2017 IEEE International Conference on Real-time Computing and Robotics (RCAR), 2017, pp. 285-290.
[45] M. Li, J. Deng, F. Zha, S. Qiu, X. Wang, and F. Chen, "Towards Online Estimation of Human Joint Muscular Torque with a Lower Limb Exoskeleton Robot," Applied Sciences, vol. 8, p. 1610, 09/11 2018.
[46] N. Anang, R. Jailani, N. Tahir, H. Manaf, and N. Mustafah, Analysis of kinematic gait parameters in chronic stroke survivors. 2016, pp. 57-62.
[47] L. Beran, P. Chmelar, and L. Rejfek, "An Increase in Estimation Accuracy Position Determination of Inertial Measurement Units," MATEC Web of Conferences, vol. 75, p. 05001, 01/01 2016.
[48] B. Hwang and D. Jeon, "A method to accurately estimate the muscular torques of human wearing exoskeletons by torque sensors," (in eng), Sensors (Basel, Switzerland), vol. 15, no. 4, pp. 8337-8357, 2015.
[49] W. T. Dempster and G. R. L. Gaughran, "Properties of body segments based on size and weight," American Journal of Anatomy, vol. 120, no. 1, pp. 33-54, 1967/01/01 1967.
[50] P. de Leva, "Adjustments to Zatsiorsky-Seluyanov's segment inertia parameters," Journal of Biomechanics, vol. 29, no. 9, pp. 1223-1230, 1996/09 1996.
[51] "單腳簡圖," 檢自:https://www.istockphoto.com/hk/vector/%E8%BA%AB%E9%AB%94%E9%83%A8%E5%88%86-%E6%A4%8D%E7%89%A9%E6%88%96%E9%9E%8B%E5%BA%95%E7%9A%84%E8%AA%AA%E6%98%8E-%E9%BB%91%E8%89%B2%E7%99%BD%E8%89%B2%E7%90%86%E6%83%B3%E7%9A%84%E7%9B%AE%E9%8C%84-%E8%B3%87%E8%A8%8A%E5%92%8C%E6%A9%9F%E6%A7%8B%E6%9D%90%E6%96%99-gm1036381124-277419168.
[52] "What Are Kneecaps For? The Importance of Mobilization," 檢自:https://www.youtube.com/watch?v=XnYO4TnpTCo&list=PLkGjXgG6qnChsxfYD3vvAxTDfIgdY-_hF&index=95&t=0s.
[53] F. Rosenblatt, "The Perceptron: A Probabilistic Model for Information Storage and Organization in The Brain," Psychological Review, pp. 65--386, 1958.
[54] D. E. Rumelhart, G. E. Hinton, and R. J. J. N. Williams, "Learning representations by back-propagating errors," vol. 323, pp. 533-536, 1986.
[55] S. Hochreiter and J. Schmidhuber, "Long Short-Term Memory," vol. 9, no. 8 %J Neural Comput., pp. 1735–1780, 1997.
[56] "股四頭肌," 檢自:https://www.sportsplanetmag.com/article/desc/17020713371556517.
[57] W. Huo, S. Mohammed, and Y. Amirat, "Impedance Reduction Control of a Knee Joint Human-Exoskeleton System," IEEE Transactions on Control Systems Technology, vol. 27, no. 6, pp. 2541-2556, 2019.