10. Nonlinear Model Predictive Control Real-Time Optimizers for Adaptive Cruise Control: A Comparative Study

Nonlinear Model Predictive Controllers are among the most popular techniques for developing intelligent control systems in automotive applications, including Adaptive Cruise Control Systems. Nonlinear Model Predictive Controllers, however, can be computationally expensive and many studies in the last decade have been dedicated to decrease their computational time. In this regard, by applying the so-called indirect optimization approach of Generalized Minimal Residual method to the Adaptive Cruise Control problem for Toyota Plug-in Hybrid Prius, this chapter contributes to the literature in three ways; first, it provides a comparative study of indirect real-time optimization methods in the context of Nonlinear Model Predictive Adaptive Cruise Control. These methods are implemented by the MPsee tool, which facilitates automatic code generation, implementation, and calibration of Nonlinear Model Predictive Controllers for MATLAB and Simulink users. Second, real-time implementability of proposed real-time optimization approaches is investigated by conducting Hardware-in-the-Loop experiments. Finally, this study argues that utilizing a Perturbed Chord modified Newton method instead of a Standard Newton method improves the computational speed up to 50%.