Smooth and Robust Path-Tracking Control for Automated Vehicles: From Theory to Real-World Application

Path tracking is a fundamental challenge in the development of automated driving systems, requiring precise control of vehicle motion while ensuring smooth and stable actuation signals. Advancements in this field often lead to increasingly complex control solutions that demand significant computational effort and are difficult to parameterize. A novel variable structure path-tracking control approach, that is based on the geometrically optimal solution of a Dubins car, offers a promising solution to this challenge. The controller generates an n-smooth and differentially bounded steering angle and, with n+1 parameters, can be tuned towards performance, robustness or low magnitude of the steering angle derivatives. In prior work, this controller demonstrated its performance, robustness and tunablity in various simulations. In this contribution, we address the challenges of implementing this controller in a real vehicle, including system dead time, low sampling rates, and discontinuous paths. Key adaptations are proposed to ensure robust performance under these conditions. The controller is integrated into a comprehensive automated driving system, incorporating planning and velocity control, and evaluated during an overtaking maneuver (double-lane change) in a real-world setting. Experimental results demonstrate good reaching and tracking behavior despite significant disturbances, with smooth steering angle outputs. This work validates the controller's practical applicability and effectiveness, highlighting its potential for real-world automated driving applications.