The presentation provides an overview of self-learning and predictive control methods relevant to the control of autonomous systems, especially robots and drones, describes the theoretical basis of the procedures, presents the current challenges and the results achieved so far.
Basics of MPC.
Learning structures (Deep NN, Gaussian process).
Control design (explicit MPC, GP-MPC, reinforcement learning).
Related directions: Increasing the robustness of nonlinear control, motion planning in a dynamically changing environment.