Lane-Keeping System with Adaptive Model Predictive Control using Online Dynamic Mode Decomposition

Autonomous vehicles have become an integral part of modern transportation sys­ tems, enabling a wide range of applications from personal mobility to industrial automation. The increasing complexity of real-world driving environments has ne­ cessitated advanced control strategies and real-time modeling techniques to ensure precision and safety. Among these challenges, lane-keeping control remains a criti­ cal focus for achieving robust and accurate trajectory tracking under dynamic and uncertain conditions.This thesis presents an integrated approach combining Online Dynamic Mode De­ composition (Online DMD) with advanced control architectures for lane-keeping ap­ plications in autonomous vehicles. The Online DMD model continuously identifies system dynamics in real time, providing the basis for adaptive control strategies. A hierarchical control framework is implemented, with upper-level proportional con­ trollers generating reference signals and lower-level controllers-Model Predictive Control (MPC) and Linear Quadratic Regulator (LQR)-executing the trajectory tracking.The proposed methodology is evaluated in a high-fidelity simulation environment using realistic road scenarios, including constant curvature roads, sharp turns, and city-like environments. External disturbances and measurement noise are incor-lll""porated to test robustness. The results demonstrate that Online DMD effectively captures system dynamics, enabling the controllers to maintain lane-keeping accu­ racy and stability. The study highlights the potential of data-driven modeling and adaptive control strategies for advancing real-world autonomous vehicle applications.