Approximate Neural Network Model for Adaptive Model Predictive Control

An approximate Neural Network (NN) model is proposed to imitate the performance of adaptive Model Predictive Control (MPC). The proposed model is used for nonlinear systems with nonlinear constraints related to automotive applications (i.e. Lane Keep Assist (LKA)). The obtained result can guarantee stability, robustness, and constraint satisfaction for the learned MPC. In this work, it is illustrated that the Approximated Adaptive Model Predictive Control (AAMPC), can reduce computational complexity and be a better solution for embedded systems.