Learning inside of model predictive control E.g. learn a neural network model of robotic dynamics: (learning) dynamics cost function sampling distribution how to optimize etc.
