Some Rules for Setting the Horizon Parameters of NMPC-Based Vehicle Path Tracking

Because nonlinear model predictive control (NMPC) can be applied to deal with system constraints explicitly and to cope with paths with curvature mutations, it has become increasingly widely used in vehicle path tracking. But up to now, the influence of horizon parameters on path tracking is not very clear. To this end, we built a controller and completed some tests using the co-simulation by MATLAB/Simulink and Carsim. According to the result, it can be known that setting the control horizon is short and the prediction horizon is similar to the control horizon or setting the control horizon is short and the prediction horizon is too long will cause the accuracy of the NMPC-based vehicle path tracking to deteriorate. As for real-time, the control horizon has a more significant impact than the prediction horizon. If the control horizon is shorter, the real-time performance is better. Further inferring, setting a shorter control horizon first, and then selecting an appropriate prediction horizon, is an appropriate way to set the horizon parameters.