The mathematical model of the controlled system is not necessary
Our technology learns from the experience
the study and use of optimal control & planning techniques applied to nonlinear dynamic systems (also valid for linear systems), both stable and unstable.
Cell-Mapping techniques combined with reinforcement learning methods. There are many examples of this kind of systems, such as aircrafts, Remotely Piloted Aircrafts Systems (RPAS), satellites, segways, etc.