This work explores the integration of process domain knowledge with online adaptation and learning methods to enhance complex control systems. It focuses on three main areas: Parameter Estimation Algorithms: A general forgetting least-squares algorithm is introduced, capable of recovering many known algorithms by specifying a weighting matrix. This algorithm combines benefits of exponential forgetting, resetting, and selective forgetting while keeping computational costs low. Adaptive Control for Solenoids: An adaptive two-degrees-of-freedom control algorithm is developed for controlling solenoid currents. This algorithm includes an adaptive feedforward controller that utilizes estimated plant parameters and model structure to improve tracking performance. The stability of the closed-loop system is proven, and the control concept is successfully applied to three different solenoids, outperforming benchmark control designs. Improving Robot Accuracy: A flexible control scheme combines a model-based controller with an online path iterative learning controller (ILC) to enhance the absolute accuracy of industrial robots. The ILC compensates for unknown residual error dynamics due to elasticity and transmission effects. Experimental validation on a 6-DoF robot shows a 95% accuracy improvement after just two trials, and the accuracy is sustained even without continuous trial-by-trial learning. Overall, this work demonstrates the significant benefits of integrating learning strategies with classical model-based control to achieve high performance in complex control systems.