This course delves into the rapidly evolving field of learning-based control, bridging fundamental concepts from machine learning and optimization with advanced control methodologies. Recognizing that modern control is inherently rooted in an optimal control framework, the course will guide participants through the journey of designing intelligent control systems that leverage data for enhanced performance and adaptability. We will begin by revisiting essential machine learning and optimization paradigms, setting the stage for understanding how these tools empower contemporary control. We will then explore both explicit data-driven control techniques and control-oriented system identification, emphasizing the principle of extracting only the necessary information for control design. Finally, the course will address challenges in numerical optimal control, focusing on strategies to accelerate solvers and formulate control problems to ensure numerical stability and computational efficiency.
