On convergence control in differential dynamic programming applied to realistic aircraft and differential game problems

Abstract

A first order algorithm for optimal control is used to solve optimal trajectories associated with aircraft applications including pursuit-evasion problems. To overcome the stability problems which occur in the algorithm, a convergence control technique is described in which a set of convergence control parameters is introduced. The effect of these parameters and how to choose them are discussed and demonstrated on optimal climb and pursuit-evasion problems. A fixed terminal time is assumed, although this is not a requirement. Numerical results are presented which illustrate the features of the algorithm, especially the ability to select different optimal solutions in the case of a nonunique solution.