Initialization of real-time optimal control

Recently real-time optimal control has attracted increasing attention since it has shown great potential in nonlinear feedback control. Computing speed and the assurance of a reliable solution in real-time have been the major limiting factors in applying real-time optimal control. In this paper we use a neighboring optimal control to generate initial guesses of the next step optimization using pseudospectral methods. In the real-time optimal control scheme we consider the subsequent optimization as the neighboring of the current optimization, the neighboring optimal control is used to generate accurate initializations for the subsequent optimization. Thus, the speed and convergence of the real-time computation can be improved with these initializations.