Nonlinear Predictive Torque Vectoring with Brake Blending for Electric Road Vehicles

The paper presents a nonlinear model predictive control algorithm for self-driving electric vehicles. The algorithm uses a nonlinear, two-track vehicle model that enables all-wheel drive torque vectoring, active front steering, and autonomous braking. The controller tracks the desired vehicle trajectory, taking into account the battery state, and enforces energy savings through regenerative braking while preventing battery overvoltage. The performance of the algorithm was tested in a high-fidelity vehicle dynamics simulator using a realistic battery model and a reference trajectory based on the Hockenheimring racetrack, showing how it increases the maximum range of the vehicle without compromising driving performance.