Robust Safety-Critical Control Framework Applied to Lane Keeping on a Real-Time Hardware-in-the-Loop Setup

Lane keeping assist system (LKAS) is one of the advanced driver assistance systems (ADAS) technologies that corrects the vehicle direction to keep it within the lane boundaries and ensures small lateral displacement even with considerable road curvature. In this work, we consider a safety-critical control framework applied to the LKAS. The driver provides a human-assisted input to achieve steering operation in the lane. If an unintentional lane departure occurs, an automatic controller acts by correcting the vehicle steering angle to respect safety constraints, expressed as control barrier functions (CBFs), that ensure the vehicle lateral displacement never exceeds predetermined bounds and keep it within the lane boundaries. The final control framework is a safety filter, or an active set invariance filter (ASIF), that filters the driver input through a quadratic programming (QP) modifying it minimally to ensure the safety constraints. Since the QP constraint depends on the system model, it is essential to account for model uncertainties and robustness to ensure that the controlled system appropriately respects safety constraints. Thus, the main contribution of this work is the proposition of a robust safety-critical control framework applied to the LKAS whose design is an extension of the procedure considered for exponential control barrier functions (ECBFs), where here we consider a sliding mode approach to deal with model uncertainties. We implement the proposed control framework in a real-time hardware-in-the-loop (HIL) setup. The results show that the safety constraints were satisfied, even with model uncertainties and considerable road curvature.