Data-driven Robust Path Tracking Control of Car-trailer Combinations Based on Dynamic Mode Decomposition

This study proposes a novel robust path tracking control (PTC) for car-trailer combinations. The motion dynamics of the car-trailer is identified using the dynamic mode decomposition with control (DMDc) algorithm. This approach does not require prior knowledge of system parameters such as mass and cornering stiffness. To account for potential modelling errors, uncertainties are considered in the identified data-driven model. Moreover, a robust H<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">∞</inf> path tracking controller is designed to ensure the system’s robustness against modelling errors and uncertainties. The path tracking controller is formulated as a linear matrix inequality (LMI) feasibility problem. Using the CarSim-Simulink platform, the motion of the autonomous car-trailer is simulated and the performance of the proposed technique is evaluated. Simulation results show that the maximum lateral and heading angle errors during the double lane change (DLC) maneuver at a speed of 50km/h are 20.08cm and 1.01deg. To further evaluate the performance of the controller, the reference path generated by experimental driving data is used for the performance evaluation of the proposed PTC. The maximum lateral and heading angle errors stay within the safe limits defined based on the road lane width.