A belief I hold strongly: The future of advanced control algorithms lies in the fusion of NMPC (Non-linear Model Predictive Control) and neural network technologies. So I want to share my toy rocket control simulation project. I’ve introduced an AI Neural Network alongside the original (nonlinear) model predictive control (NMPC). It’s a bit of a trade-off: the AI runs faster and fits easier on target platforms, though it’s hard to beat the NMPC in the simulation. Full source code available and maybe interesting for others as well.