Robust Path Tracking Control of Self-driving Vehicles with Measurement Deficiencies and Parametric Uncertainty

This study presents a novel approach for path tracking of Self-driving Vehicles (SDVs) with measurement deficiencies in sensor outputs and uncertainties in tyre cornering stiffness. The unavailability and inaccuracy of measured signals in SDVs are unavoidable due to practical factors such as sensor failure and reception blockage. Furthermore, the probability of deficiencies in measured signals may vary in different sensors. This study models the sensors’ measurement deficiency using the fading measurement phenomenon represented by independent random variables. Norm-bounded uncertainties are also considered to model uncertain time-varying tyres’ cornering stiffness. A path tracker is designed using the Lyapunov stability theorem to ensure vehicle stability. The H_∞ H ∞ performance index is fulfilled in the presence of measurement deficiencies and parametric uncertainties. The controller’s desired gain is determined by solving an optimisation problem based on a Linear Matrix Inequality (LMI). A Double-lane Change (DLC) manoeuvre, defined based on ISO 3888-2 with Moose test speed requirements, and a Single-lane Change (SLC) manoeuvre with a speed of 90 kmh ⁻¹ - 1 are used for simulation and evaluation of the efficacy of the proposed controller. The simulations are performed in the CarSim–Simulink co-simulation platform. The results show that the proposed controller produces acceptable lateral and heading angle errors when the average probability of missing measurement in individual and multiple states is less than $$30 %$$ 30 % . Similar results are observed in the frequency content of the steering angle and the lateral acceleration of the vehicle. The overall outcomes of the work reflect the robustness and performance of the proposed method in handling uncertainties and measurement deficiencies.