Design and implementation of a torque vectoring control for Formula Student Driverless application = NON APPLICABILE

This thesis work is focused on the design of a Torque Vectoring (TV) architecture for a 4WD (Four Wheel Drive) electrical racing, in the context of the participation by the Polytechnic of Turin in the Formula Student Driverless competition organized by the Society of Automotive Engineers (SAE). The TV aims to optimally distribute the total required torque, both in traction and in braking, in order to produce an additional moment acting on the vehicle’s yaw rate. The control TV architecture is mainly composed of three sub-systems: a Reference Generator, a High-Level Controller based on a Linear Quadratic Regulator (LQR) and a Low-Level Controller for the motor torque allocation to the wheels, solving a Quadratic Programming (QP) optimization problem. This project is based on a co-simulation between Vi-CarRealTime and MATLAB/Simulink. The former provides an accurate full Degree of Freedom model of the vehicle and represent the plant of the whole control system; the latter, instead, is actually used for the development of the Torque Vectoring controller and the simulation set up. Due to the autonomous feature of the prototype, the simulations have been conducted generating a pre-planned path by MATLAB script and employing a Model Predictive Controller (MPC) for guaranteeing the path tracking. The developed Torque Vectoring aims to control the lateral dynamics of the prototype to enhance its cornering performances by the allocation of a suitable torque to each in-wheel electric motor. A remarkable consideration high-lighted by the results is the advantage of the TV regarding the MPC, since the TV acts as a model following to force the predictive model to work more accurately and closer to the plant model, preventing model mismatches due to non-linearities, approximations and eventual inexact parameters estimations.