Framework for reproducible validation of a real-time capable vehicle lateral stability controller

Following the rise in technological advancements, in recent years the idea of mobility has undergone significant transformation. The growing tendency towards electrification has prompted innovative control strategies, which in turn have introduced new challenges. Current research goals in the automotive field may be summarised in two categories: ensuring passenger safety and enhance the robustness and reliability of such technologies. Within this framework, driver assistance systems like the Electronic Stability Control (ESC) are effective in avoiding potential loss of lateral stability. Extensive efforts have been made in the literature on the topic to refine control algorithms of this sort, but most of them show little tendency to progress beyond theoretical definition or software simulations. Despite such efforts, a staggering number of annual casualties is still registered, which calls for an enhancement of stabilization control strategies. This thesis proposes a rigorous validation workflow for an individual-wheel-torquebased vehicle stability controller, aimed at establishing a robust and replicable testing protocol for a generic lateral stability controller to run in real time. The sequence of progressive validation steps starts with an initial design phase, where the controller is formulated according to specific performance requirements. It then envisions an offline co-simulation phase, where the controller undergoes preliminary testing in the form of software, running concurrently with a suitable vehicle testbench. Later, the controller and vehicle model are moved to a software-independent real-time-enabled testing platform in the so-called real-time co-simulation phase. Finally, the controller is turned into code and deployed to its target hardware (Hardware-in-the-loop), ultimately enabling full-scale vehicle validation. Further noteworthy contributions include a novel strategy for vehicle reference behavior design, an innovative method to estimate what is arguably the strongest vehicle stability indicator, i.e. the sideslip angle, using both a machine learning-based and a parametric approach, and the analytical assessment of the vehicle stability region to reliably assess the operating condition of the vehicle.