Comparative Analysis of Nonlinear MPC Formulations for Constrained Vehicle Path Following with Model Mismatch

This Paper implements and compares four advanced nonlinear Model Predictive Control (MPC) strategies—sequential linearization with Quadratic Programming (QP), MPC with Nonlinear Prediction and Linearization along the Trajectory (MPC-NLPT), Interior-Point (IP) nonlinear MPC, and Sequential Quadratic Programming (SQP) nonlinear MPC—for trajectory tracking on the same nonlinear plant used in prior mini-projects. All methods are benchmarked against a baseline infinite-horizon Linear Quadratic (LQ) controller. Simulation results demonstrate that the nonlinear approaches achieve superior tracking accuracy (up to ۴۵ % lower IAE and reduced maximum deviation) and better constraint satisfaction than the LQ controller, while QP-based methods offer the best computational speed. The study highlights the fundamental trade-off between control precision and real-time feasibility in nonlinear MPC.

کلیدواژه ها:

predictive control ، autonomous racing car ، nonlinear control ، path tracking