七軸機器手臂模型建立及疊代學習控制之應用

Abstract

傳統六軸機器手臂已發展成熟並廣泛應用在加工製造業上，然而具冗餘自由度的七軸機器手臂仍有發展空間且有研究價值。本論文中的研究對象為一七軸S-R-S type仿生機器手臂，其架構為模仿人類手臂構造，因此能體現出其接近於人類的功能性。對於此架構之機器手臂，首先需對其建構運動學模型與動力學模型，由於機器手臂具冗餘自由度，因此需要透過參數化方法解決逆運動學無限組解之問題。另外，在尋求機器手臂奇異點代數解時，透過將參考點由手臂末端點移置腕部位置簡化求解複雜度。對於機器手臂動力學模型，分別採用Lagrangian dynamic equation及Iterative Newton-Euler dynamic equation求解並交互驗證，將結果與馬達動態模型結合得到馬達力矩與機器手臂各軸角度之關係。由於動力學模型的複雜度高，因此本論文提出透過各軸設置煞車，對各軸獨立做動力學參數估測之流程。對於機器手臂控制方法，本論文結合計算力矩控制器、數位擾動估測器、疊代學習控制器三種控制方法，透過計算力矩控制器使機器手臂各軸成為獨立之線性系統，使用數位擾動估測器增加系統強建性並消除非重複性擾動，透過疊代學習控制將重複性擾動消除增加機器手臂系統效能。此外，針對力矩型與參考型疊代學習控制兩種架構進行穩定性分析並比較其中差異。

Traditional 6-degree-of-freedom (DOF) robots have already been developed and applied in industry for a long time. However, 7-DOF redundant robots exhibit many challenging open questions which are worthy of further researches. This thesis focuses on a 7-DOF S-R-S type biomorphic robot. Because its architecture is similar to a human’s arm, it can do almost all works as human does. In the first step of this research, we need to develop the kinematic and dynamic models of the robot. Because the robot has a redundant joint, we need to parameterize the infinitely many solutions to the inverse kinematics. To solve the singularity problem, the wrist instead of the end effector is chosen as the reference point to make the singularity problem tractable. To establish the robot dynamic model, we use both the Lagrangian and Newton-Euler dynamic equations for cross validation. However, the dynamic model is too complicated to identify all parameters in one experiment. Therefore this thesis presents a procedure to reduce the model complexity by estimating the model parameters joint by joint. For the aspect of robot control, this thesis proposes a method combing computed torque control, digital disturbance observer and iterative learning control (ILC). Two ILC structures, i.e. torque ILC and reference ILC, are studied. Then simulations are conducted to verify and compare the performance of the proposed method.