Abstract
A finite-horizon two-player non-zero-sum differential game with linear dynamics is considered. For each player, there is its own quadratic cost functional to be minimized. The game is analyzed with respect to the state-feedback Nash equilibrium. Due to the well-known classical results [2, 20], the solution of such a game is reduced to the solution of the terminal-value problem for the set of two coupled non-linear (Riccati-type) matrix differential equations in the entire time interval of the game’s duration. Unfortunately, this problem may have a conjugate point, meaning the non-existence of its solution in the entire time interval of the game’s duration. By a proper version of the artificial parameter method, a sufficient condition for the existence of the solution to this terminal-value problem in the entire time interval is derived. Directly based on the game’s data, the established condition is easily verifiable. The fulfillment of this condition yields the upper bound of the length of the aforementioned time interval which can be, in general, sufficiently small depending on the data of the game. Using this condition, an approximate analytical solution to the aforementioned terminal-value problem is obtained as a partial sum of some infinite series of functions arising in the method of artificial parameter. Obtaining each addend of this partial sum requires only an integration of a known matrix-valued function of time. This approximate solution is further used to design an approximate Nash equilibrium component of each player and to design an approximate Nash equilibrium pair of the players’ controls in the considered game. The theoretical results of the paper are applied to approximate the solution of a problem of pursuit-evasion engagement between two flying vehicles.

This publication has 12 references indexed in Scilit: