Predictive Torque Vectoring Vehicle Control Based on a Linear Time Varying Model

In this paper we propose a torque vectoring and steering system based on a model predictive control (MPC) algorithm for a linear time varying (LTV) vehicle model. The goal is to ensure good reference tracking, while fulfilling various physical and design constraints. Non-linear vehicle model uses steering angle and wheel slip ratios as inputs. For control purposes, the model is linearized along the trajectory and then transformed to discrete time domain. Afterwards, the controller is tested using the sine with dwell driving manoeuvre with different prediction horizon lengths and different linearization approaches. Additionally, autonomously steered vehicle is compared to semi-autonomous vehicle in which the steering angle is determined by the driver and the results are discussed.