A path tracking approach for autonomous driving on slippery surfaces

In this paper a linear Model Predictive Control (MPC) based controller for the path tracking of a given trajectory is proposed, which generates suitable references for the dynamical vehicle states. The goal of the path tracking is to achieve good performance on slippery conditions. The equations describing the evolution of the path are linearized in order to obtain a suitable model, and linear constraints on the produced reference states are applied based on the prior knowledge of the reachable equilibrium states which characterize steady-state cornering. A previously proposed low-level LQR-ISM (Integral Sliding Mode) controller is exploited for reference tracking, where the ISM component is required in order to enhance the robustness with respect to the several approximations introduced by the linearized models. IPG-CarMaker, a high-fidelity vehicle dynamics simulation software, is used for the validation. The performance is verified on a wet asphalt track characterized by segments of different curvature, and compared against the CarMaker driver.