請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/2303
標題: | 使用低成本雷射測距儀與即時組織硬度估測之適應性軌跡規劃系統於機器人按摩之研究 Adaptive Trajectory Generation of Robotics Therapeutic Massage Using Low-Cost Laser Range Finder with In-Situ Tissue Stiffness Estimation |
作者: | Li-Tung Tang 唐莉彤 |
指導教授: | 羅仁權(Ren C. Luo) |
關鍵字: | 雙手臂機器人,低成本雷射測距儀,適應性按摩軌跡規劃,即時組織硬度估測, dual arm robot,low-cost laser range finder,adaptive massage trajectory generation,in-situ tissue stiffness estimation, |
出版年 : | 2017 |
學位: | 碩士 |
摘要: | 隨著人口老化與人們對生活品質的要求提升,服務與醫療照護型機器人逐漸走入我們的生活中,具有相當大的市場。現代人壓力大作息不正常,常常造成肩背痠痛,而按摩為一種常見的舒緩方式,其不只幫助人們放鬆也可被視為是一種治療。若機器人可被應用於按摩上,不僅能解決醫療人力不足的問題也可增進醫療品質。
本篇論文以雙手臂機器人於肩背按摩為主軸,發展出一套完整的自動化按摩流程。該流程包含使用自製的低成本雷射測距儀產生專屬個人的適應性按摩軌跡,以及加入即時組織估測系統使機器人在按摩過程中能自行估測被按摩部位組織的厚度與硬度。按摩前置部分。在以往,機器人的按摩軌跡大多透過教導的方式或是利用使用者介面來產生,此作法耗時又耗費人力。而在此我們使用自製低成本的3D雷射測距儀搭配現有的變形演算法找出人體肩部與背部的按摩軌跡。該雷射測距儀具有與市售產品相當的精度且售價只需一般產品十分之一不到的價格。我們利用此雷射測距儀掃描人體以產生背部的立體模型,並將該模型與預先建立已具有標準按摩軌跡的模型做映射來找出專屬於每個人的按摩軌跡。實驗結果顯示此方法所找出的按摩指壓點與專業按摩師所認可的指壓點平均距離誤差為1公分,產生軌跡的的時間也只需教導方式的一半。在按摩的過程中,則加入了即時組織估測系統。我們透過7軸手臂等速按壓人體背部找出皮膚組織力量與深度的曲線,將人體的皮膚特性模組化為一正切函數。實驗中機器人雙手前端裝有力/力矩感測器,可即時量測按摩器具與被按摩者間所受的力。我們將此力與機器人末端的移動做為估測演算法中訊號的輸出與輸入,並藉由遞迴最小平方法加上遺忘因子在按摩中即時找出函數當中的參數以估測組織的軟硬。該實驗分為模擬與實際量測。模擬結果顯示該演算法可在20秒內估測出組織的特性。而在實際量測中,機器人對不同體型的受測者進行按壓按摩並估測骨頭周遭、硬肌肉、軟肌肉的組織的特性。實驗結果顯示其可在1分鐘內估測出組織的特性且其結果與先前模組化的軟組織模型趨勢相同並有潛力可做為判斷被按摩者身體狀況的標準。 With the increasing rate of the elderly and the promotion of life quality, service robot and health care robot play important roles in our life and have huge market potential. Modern people usually suffer shoulder and back pain because of the pressure and irregular routine. Massage, a common way to alleviate the pain, can help people relax as a therapy. If robots are able to be supplied for massage, it can solve the shortage of medical human and promote the quality of therapy. In this thesis, we develop a complete automated procedure on dual arm robot for shoulder and back massage application. It contains using self-developed low-cost laser range finder (LRF) to generate adaptive massage trajectory which is individual person specific and implementing in-situ tissue stiffness estimation to make robot can figure out the characteristics of patient’s soft tissue during massage. In pre-massage procedure, prior researches often apply teach and play or user interface to generating massage trajectory and these techniques are time-consuming and laborious. Here we use a 3D low-cost LRF made by ourselves and existing morphing algorithm to find the trajectory of shoulder and back massage. This LRF has the comparable performance as commercial products and its cost is less than 10% of the cost of general products. We use the LRF to scan human’s back and generate a 3D model. Then the standard model which has the massage trajectory we decide in advance will be mapped on the model of human’s back to generate the massage trajectory for different people with different size. The experimental results demonstrate that the average distance error between the acupressure points indicated by our method and the acupressure points defined by the massage therapist is 1cm and our method can save more than half the time needed with teach and play. During the massage, we add an in-situ tissue stiffness estimation. We use a 7 DoF arm to find the indentation-force curve of soft tissue and model the characteristics of soft tissue as a tangent function. The wrists of the robot are equipped with force/torque sensors which can detect the force between massage tools and the patient. We take this force and the movement of robot manipulator as output and input signals and apply the recursive modified least-squares algorithm with forgetting factor to find the parameters of the function and estimate the hardness and thickness of tissue. From simulation results, the algorithm can estimate the characteristics of soft tissue in 20s. In the experiment, the robot executes pressing massage on different subjects with different body size and finds the characteristics of three types of soft tissues, which are the tissue that covers bony area, hard muscle, and soft muscle. From experimental results, it can estimate the characteristics of soft tissue in 1 minute. The estimation results have the same trend as the model of soft tissue we build previously and have the potential to be the judging criteria of body condition. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/2303 |
DOI: | 10.6342/NTU201703367 |
全文授權: | 同意授權(全球公開) |
顯示於系所單位: | 電機工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-106-1.pdf | 6.12 MB | Adobe PDF | 檢視/開啟 |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。