Akademik Veri Yönetim Sistemi
Karaelmas Fen ve Mühendislik Dergisi, cilt.15, sa.3, ss.56-73, 2025 (TRDizin)
This paper examines the challenges associated with path tracking in autonomous vehicles, employing Model Predictive Control (MPC) based on a single-track vehicle model. An MPC controller is developed based on the state-space representation of this vehicle model. Longitudinal speed is constant, and the algorithm's performance is tested with different vehicle longitudinal speeds. The generated reference path is sent to the algorithm as a reference lateral position, and the MPC controller predicts the future parameters to correctly track these given reference paths. The Proportional-Integral (PI) control method also tested on model and it is compared with MPC. The model is tested with two different paths, sinusoidal and F1 Barcelona circuit paths. It has also been tested with five different simulation scenarios. These scenarios examine the effects of parameters such as prediction horizon, control horizon, weight matrices, longitudinal velocity, computation time, and control methods. MPC optimization algorithms and simulation results are realized on the MATLAB/Simulink environment using CasADI, an open-source third-party library for nonlinear optimization and solving nonlinear equations.