Adaptive Robust Controller for handling Unknown Uncertainty of Robotic Manipulators

This thesis introduces a new adaptive robust control scheme, specially designed to improve the stability of robotic manipulators and the accuracy of their performance, whenever it takes place under uncertain and dynamic circumstances. Classical control approaches have an acceptable performance in structured environments, however, they usually have problems with model inaccuracy, tolerating external disturbances, and real-time parameter variations. In order to solve these problems, this research has produced a robust Adaptive Feedback Linearization Controller (ARFLC) which is able to make real-time correction of both structured and unstructured uncertainties. The approach proposed in this study is to apply the method to two benchmark robotic systems, namely a 2-DOF RR manipulator and a 4-DOF SCARA manipulator. The construction of kinematic and dynamic models of both robots which includes the Denavit-Hartenberg parameters and Lagrangian formulation has been conducted. The controller contains an adaptive gain feature, whereby the adaptation coefficient is updated selectively, triggered only when the system faces atypical grinds in a particular way. This kind of pure adaptation of functions does not merely contribute to computational efficiency but also prevents overcompensing due to minor disturbances. Simulation results confirmed the supremacy of the proposed ARFLC approach the problems of trajectory tracking, robustness, and speed of convergence, as compared to traditional ones. The results get to the root of the issue and confirm the use of adaptive robust control as one of the most practical ways in modern-day robot applications, where there is always the chance of uncertainty and the need for real-time adaptability is key.