Data science for decision aiding UAV control

The dynamics of a human operated nonlinear unmanned aerial vehicle (NUAV) with a given controller at the inner loop is governed by the admissible nonzero initial conditions and the pilot inputs. The movements of the joystick connected to the throttle and the control surfaces in a way are the indications of the pilot decision points to operate the stabilized NUAV. If these decisions fall short to avoid an obstacle or a denied air space, data science to generate the admissible control inputs through appropriate initial conditions becomes an important problem. Such control inputs at the autonomous decision points (decisions not made by the human pilot) complement the pilot inputs and may be used to avoid the obstacles and the denied airspace. In this paper, the rules to choose these control inputs at the autonomous decision points are discussed. A three degree of freedom aircraft in pitch plane is considered for illustrations.