基於使用者意圖之行動輔具設計

Abstract

近年來，醫療技術的革新為人類帶來了較長的壽命，妥善照顧這群日益龐大的銀髮族群成了社會上很重要的一項議題，在老人常見的問題中又以行動不便的問題最常遇到，因此在本論文中，我們設計一能有效辨識使用者意圖的力感握把，將其配置於命名為i-go的被動式機器人行動輔具，利用i-go幫助銀髮族穩健行走，並藉由此握把有效辨識使用者意圖，以降低銀髮族行走時跌倒的可能性。本文也提出了使用者意圖辨識演算法，藉由Lasso模型推論出使用者意圖與實際推拉力、旋轉力矩的關係，並通過PCA演算法得出各種意圖的權重值，在銀髮族施加各種意圖時，可以該意圖的權重值及本論文提出的模糊控制演算法，控制i-go選擇相對應的煞車力輸出。此外，為了因應個別使用者操作習慣的不同，使用了ANFIS演算法改善我們設計的模糊控制器以適應不同的使用者，達到更精確的煞車力輔助。

In recent years, the progress in medical innovation has contributed to longer life-span for human being. How to take good care of the large amount of the elderly now becomes an important social issue. Among the general health problems for the elderly, disabilities surge as major issues to face. Hence, in this thesis, upon a robot walking helper developed in our laboratory, we attach force-sensing grip handle which can be used to effectively recognize the intention of the users. We use Lasso model for user intention recognition that we can infer the relationship between users' intention and measured pushing-pulling force with rotating torque. We also use the algorithm of PCA to get every intension’s weight. When the aged use different intention, they can use this intension’s weight and the algorithm of fuzzy to control the i-go to choose corresponding output of braking force. Besides, to adapt ever user’s different habits, we apply the algorithm of ANFIS to improve self-designed fuzzy controller to adapt to different users in achieving accurate assistance in braking force.