由於機器手臂可以不間斷地持續工作且擁有高準確性的效能，因此在現今自動化工廠中它通常扮演著非常重要的角色，然而機器手臂的移動軌跡的追蹤過程中，時常會受到來自於環境和未知的外部干擾以及各種參數不確定性等影響，導致控制器無法精確地追蹤預設的期望軌跡。有鑒於此，本篇論文提出了一種新型強健控制方法，它是基於一個總干擾及不確定性的觀測器同時結合先進控制技術，應用於機器手臂使其同時兼具高強健與高精準之控制要求。首先，控制器整合一個動量觀測器與二階滑模技術，使它可以擁有比傳統動量觀測器更強的追蹤效果，以至於補償回控制器後後能使在環境干擾及參數不確定性下有很好的誤差收斂結果。另外，我們也設計了一個前饋控制器和採用自適應律法則來即時調整超螺旋滑模控制器的參數大小，使其所提出的控制器能擁有更快的收斂速度以及更小的穩態追蹤誤差。其次，我們使用Lyapunov方法來分析整體控制器的穩定性，藉由理論推導來證明所提出的控制器可以在有限時間內收斂。最後，以一系列的數值模擬與實驗結果來驗證所提出控制方法，我們也同時比較傳統的動量觀測器以及前饋超螺旋滑模控制器，並且在真實的二軸機器手臂上進行實驗證明與效能比較。;Robot manipulators are widely used in modern automated factories due to their ability to operate continuously with high precision. However, when executing a desired trajectory, they are often subjected to external disturbances from the working environment as well as parametric uncertainties. These factors can degrade tracking performance and prevent the manipulator from accurately following the predefined trajectory. To overcome these challenges, this thesis proposes a novel robust control strategy that integrates a lumped uncertainty observer with advanced control techniques. The proposed method is applied to robot manipulators to achieve both high robustness and high-precision trajectory tracking under uncertain conditions. First, a momentum observer combined with second-order sliding mode technique is designed. This observer provides more accurate disturbance estimation compared to traditional momentum observers. By feeding the estimated uncertainties back into the controller, the system achieves improved error convergence even under parametric uncertainties and external disturbances. In addition, we also designed an adaptive law to dynamically adjust the parameters of the super-twisting sliding mode controller, which is combined with a feedforward controller. The proposed control method achieves faster convergence and reduced steady-state tracking errors. Secondly, Lyapunov theory is employed to analyze the stability of both the observer and the controller. Through theoretical derivation, it is proven that the proposed control method ensures finite-time convergence. Finally, a series of numerical simulations and experimental validations are conducted to verify the effectiveness of the proposed control method. The performance is compared with that of the traditional momentum observer and the feedforward super-twisting sliding mode controller. Experimental verification and performance comparison are carried out on a real two-link robotic manipulator.
