A coevolutionary minimax solver and its application to autopilot design

Abstract

Traditional numerical solvers for minimax problems are calculus based. In this paper, we propose an evolutionary computation based numerical solver for minimax problems. We view the minimax problem as a game between two players with opposite objectives. One player trys to minimize the cost and the other trys to maximize it where strategies are expressed as populations. The strategies from each players are matched and scored. Only the best strategies will survive for each players. The surviving strategies will make offsprings for the next generation. We also consider relative evaluation method for faster convergence and number of matches between the opponents. The proposed technique is applied to a missile acceleration control problem for verification.

A co-evolutionary method for pursuit-evasion games with non-zero lethal radii

“…The conventional co-evolutionary algorithm introduced by Park and Tahk [6] is a stochastic global search algorithm for obtaining a saddle point of a minimax parameter optimization problem. If a minimax problem has a saddle point solution, then it can be treated as a zero-sum game.…”

“…The co-evolutionary algorithm [6] can also be a good candidate in solving pursuit-evasion games. It was developed as a minimax solver for saddle point optimization problems; it has provided remarkable results when applied to the robust control of an aircraft [6], and a constrained parameter optimization [7].…”

“…The co-evolutionary algorithm [6] can also be a good candidate in solving pursuit-evasion games. It was developed as a minimax solver for saddle point optimization problems; it has provided remarkable results when applied to the robust control of an aircraft [6], and a constrained parameter optimization [7]. Since a direct approach to pursuit-evasion game leads to a saddle point parameter optimization, the basic architecture of the co-evolutionary algorithm is very suitable for solving pursuit-evasion games.…”