Near-Optimal Evasion from Acceleration Bounded Modern Pursuers

Target evasion from modern missiles is a very challenging task because modern interception missiles usually have substantial agility and maneuver advantages over the target. This paper proposes novel near-optimal evasion strategies from any linear missile guidance law that exploit the pursuer’s two main weaknesses: the inherent time delay of the evader’s acceleration estimate and the pursuer’s acceleration bound. In the derivation, the pursuer and the evader are assumed to have arbitrary-order linear dynamics with bounded acceleration commands. The pursuer is assumed to have a delayed evader acceleration estimation, and the evader is assumed to know the pursuer’s guidance law. The problem is posed as a bounded optimal control problem with the miss distance as the performance index, and the necessary analytical optimality conditions are derived. The problem is then solved iteratively using backward and forward propagation of the costate and state dynamics until the solution converges. Furthermore, an additional gradient-descent-based algorithm is derived to improve the performance and robustness of the solution. The evasion strategies are extensively evaluated in deterministic and stochastic Monte Carlo simulations. It is shown that the proposed evasion strategies have substantially better evasion performance than state-of-the-art evasion strategies that only exploit the pursuer’s estimation delay.