Comparison of a fast-nonlinear model predictive vs an error-cubic proportional derivative torque vectoring controller by a hexapod driving simulator

Model predictive control (MPC) exhibits a big potential thanks to its predictive capabilities for various fields such as automated driving or chassis active control systems. Its application has been extended to torque vectoring systems which help to operate the vehicle at the limit handling region. However, tyre formulation, which is included on the MPC internal model, is usually simplified in order to gain computational speed. Besides, evaluation methods are typically performed using driver models and simulation, but rarely by employing a high-fidelity driving simulator and track model. In this paper a new nonlinear model predictive torque vectoring controller is presented which includes the complete tyre set of equations, and computational speed is increased thanks to a control horizon polynomial fitting, in order to enable a real time operation of both controller and complete vehicle simulation on a driving simulator. Moreover, the performance of the nonlinear MPC is compared with an error cube proportional derivative controller, which has previously demonstrated excellent performance on these kinds of scenarios. Results show that, even though the fast nonlinear MPC shows a higher control actuation, the error cube proportional derivative controller achieves a better lap time and a similar yaw rate tracking performance. Results files coming out from this research are shared hereby.